- Email: [email protected]
Fatigue resistance of an aluminium one-component polyurethane adhesive joint for the automotive industry: Effect of surface roughness and adhesive thickness
Author’s Accepted Manuscript Fatigue resistance of an aluminium one-component polyurethane adhesive joint for the automotive industry: Effect of surface roughness and adhesive thickness Y. Boutar, S. Naïmi, S. Mezlini, R.J.C. Carbas, L.F.M. da Silva, M. Ben Sik Ali www.elsevier.com/locate/ijadhadh
PII: DOI: Reference:
S0143-7496(18)30045-9 https://doi.org/10.1016/j.ijadhadh.2018.02.012 JAAD2137
To appear in: International Journal of Adhesion and Adhesives Cite this article as: Y. Boutar, S. Naïmi, S. Mezlini, R.J.C. Carbas, L.F.M. da Silva and M. Ben Sik Ali, Fatigue resistance of an aluminium one-component polyurethane adhesive joint for the automotive industry: Effect of surface roughness and adhesive thickness, International Journal of Adhesion and Adhesives, https://doi.org/10.1016/j.ijadhadh.2018.02.012 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Fatigue resistance of an aluminium one-component polyurethane adhesive joint for the automotive industry: Effect of surface roughness and adhesive thickness Y. Boutar 1*, S. Naïmi 1, S. Mezlini 1, R.J.C. Carbas2, L.F.M. da Silva2, M. Ben Sik Ali3 1
Mechanical Engineering Laboratory, National Engineering School of Monastir, Monastir University, Monastir, Tunisia 2
Faculdade de Engenharia, Departamento de Engenharia Mecânica, Universidade do Porto, Porto, Portugal 3
*
Automotive Industry, ICAR, Sousse, Tunisia
Corresponding author, Email:[email protected]
Abstract Fatigue is a crucial type of loading for many structural components that contain adhesive bonding systems such as the automotive industry. One of the main advantages of adhesives is their weight-reduction aptitudes. Furthermore, adhesively bonded joints allow a good damping to the fatigue's solicitation with fewer sources of stress concentration. In addition, a suitable surface treatment allows a great joint interface adhesion under cyclic loading. Hence, the mechanical preparation of the bonded surface influences directly the bonded joint performance. Therefore, a careful consideration should be given to the choice of the bonded adhesive thickness. The aim of this paper is to investigate the influence of surface roughness of an aluminium alloy and the adhesive thickness of a one-component polyurethane adhesive, to be used in several parts for buses structure, on the fatigue behaviour of single lap joints. To achieve this purpose, four different surface roughnesses were prepared on aluminium specimens, with an arithmetic average height from Ra ≈ 0.6 µm to Ra ≈ 1.5 µm. Bonded specimens with four adhesive 1
thicknesses from 0.3 mm to 2 mm were manufactured and tested in dynamic tests. The results confirm that there is an optimum combination of surface roughness and adhesive thickness providing the maximum fatigue life. Moreover, good agreement was found between this investigation and a previous work using the same parameters under static loading.
Keywords Single lap joint; Polyurethane adhesive; Surface roughness; Adhesive thickness; Fatigue life.
1. Introduction The use of adhesive bonding in the industry has greatly increased in recent years. However, its use in truly structural applications is still often limited. This is mainly due to a lack of confidence in the performance of adhesive joints, since the mechanical performance of the joints may deteriorate upon being subjected to cyclic-fatigue loading [1]. Therefore, an understanding of the effect of substrate's surface roughness and bondline thickness on the cyclic loading behaviour of adhesive joints can help to select an accurate surface treatment and an optimal adhesive thickness. Thus, the ability to quantitatively describe the reduction in performance and to predict the lifetime of bonded joints would be a powerful tool. This enables manufacturers to make wider and more efficient use of adhesive bonding.
In that context, there has been many papers dealing with the effect of bondline thickness and surface treatments under cyclic loading. Mazumdar and Mallick [2] discuss the static
2
and fatigue behaviour of adhesively bonded single lap joints in Sheet Molding Compound SMC-SMC composites. By studying the effects of lap length and adhesive thickness, the authors show that these two parameters have a negligible effect on the ratio of fatigue strength to static strength. The fatigue strength at 106 cycles is approximately 50% to 54% of the static strength for various adhesive thicknesses and lap lengths investigated in this study.
Moreover, in order to predict the cyclic-fatigue performance of single-lap joints and a typical bonded component, a fracture-mechanics approach was used by Curley et al. [3]. In their work, mild-steel substrates and a rubber-toughened hot-curing epoxy adhesive were used. For lap joints subjected to the ‘wet’ fatigue tests, the joints visually exhibited an apparent interfacial failure. But, for the tests undertaken in the ‘dry’ environment, failure occurred mainly in the adhesive layer, but very close to the interface. The researchers prove that the results from the fracture-mechanics tests not only provided the experimental data needed for the modelling studies but also indicated the expected failure path for the lap joints, which greatly assisted in the detailed modelling of the location of the cracks in these joints. Rushforth et al. [4] investigated the effect of surface pretreatment and moisture on the fatigue performance of adhesively-bonded aluminium. Single lap joints (SLJs) were tested to failure under fatigue loading to ascertain the effects of surface pre-treatment and moisture in both durability and fracture path. The aluminium substrate used was AA5754 with and without a silicon-based pre-treatment. The fatigue performance of pretreated and untreated joints was similar when tested in air, but the performance of the untreated joints was greatly reduced when tested in 96% relative humidity (r.h.) and temperature at 26±1°C. The differences in fatigue 3
performance between untreated and pretreated joints in 96% (r.h.) are thought to be due to differences in the failure mode of the joints arising from the improved interfacial bonding between the pre-treatment and the adhesive. Pretreated joints failed predominantly cohesively through the adhesive, while untreated joints failed mainly interfacially at the adhesive/aluminium interface. Abel et al. [5] proposed an accelerated test method, based upon cyclic-fatigue testing using a fracture-mechanics approach, to study the durability of organosilane-pretreated joints which were adhesively-bonded using a hot-cured epoxy-film adhesive. The silane primer used as a pre-treatment was shown to be effective in increasing the joint durability, compared with a simple grit-blast and degrease (GBD) treatment and, indeed, gives comparable results to using a chromicacid etch (CAE) pre-treatment. Furthermore, da Silva et al.[6] studied single lap joints with patterns under cyclic loading with and without chemical surface treatment and bonded with a brittle adhesive. Fatigue testing showed that the surface patterns have a beneficial effect on the behaviour of the specimens under cyclic conditions. At the same load, patterned specimens consistently endured more cycles than the specimens with no pattern, when no additional chemical surface treatment was applied.
Likewise, there are several papers that deal with the effect of bondline thickness or both surface treatment and adhesive thickness on the behaviour of cycle fatigue tests for bonded joint. In the paper of Abou-Hamad et al. [7], an experimental method based on fatigue crack growth of an aluminium bonded DCB specimens was adopted. A twocomponent epoxy was used. Three bondline thickness was investigated, namely: 0.3, 0.8 and 1 mm. The results show that the increase of adhesive thickness has a significant effect on fatigue crack growth. The larger the bond line thickness, the larger is the fatigue 4
crack growth resistance. Banea et al. [8], examined the mechanical properties of room temperature vulcanising (RTV) silicone adhesives. SLJs were fabricated and tested to assess the adhesive performance in a joint. The influence of adhesive thickness (0.5 and 1 mm) on the fatigue behaviour was made. The fatigue tests on RTV silicone adhesives SLJs showed a fatigue limit of approximately 30% of the static failure load. Moreover, Abdel Wahab [9] and Costa et al. [10,11], showed that one of the most parameters that affect fatigue strength and lifetime of bonded assembly is the surface treatment of the adherends. Therefore, these parameters need to be optimised to guarantee the bonded assembly durability. Recently, de Barros et al. [12] analysed the changes in fatigue resistance due to surface preparation in terms of a total number of cycles, further confirming that the adherends surfaces need to be carefully prepared prior to bonding.
However, modern polyurethane adhesives are increasingly being used in structural joining applications as they present important advantages in terms of damping, impact, fatigue, and safety which are critical factors in the automotive industry [13]. On the other hand, most of the results from the literature regarding the effect of surface treatment and adhesive thickness are for typical structural epoxy adhesives, which are generally formulated to perform in thin sections, while polyurethane adhesives are designed to perform in thicker sections and might have a different behaviour as a function of the design parameters. Therefore, it is important to know how polyurethane adhesive-bonded joints performance is affected by the geometric parameters. A recent static study by Boutar et al. [14,15], on the effect of surface roughness and adhesive thickness for an aluminium one-component polyurethane bonded assembly, show that there is an optimum couple of parameter allowing the highest shear strength for this joints. Hence, it 5
is worthy to distinguish how polyurethane adhesive-bonded joints performance is affected by surface roughness and adhesive thickness when they are subjected to cyclic fatigue loading.
In this study, four different surface roughness were prepared on aluminium specimens, with an arithmetic average height from Ra ≈ 0.6 µm to Ra ≈ 1.5 µm. Bonded SLJs specimens with four adhesive thicknesses from 0.3 mm to 2 mm were manufactured and tested in dynamic tests. During tests, the number of life cycles was recorded. Fracture surfaces were examined visually and using high-powered microscope SEM and EDS so that a detailed picture of the failure mechanism could be obtained.
2. Experimental details 2.1. Material The materials used in this study were an aluminium–copper alloy 5754 (see Table 1) and a one-component polyurethane adhesive DINITROL 500. Indeed, The Dinitrol 500 adhesive specimens were tested at Room Temperature in tension using a testing machine INSTRON model 3367, under a constant crosshead rate of 5mm/min. Three ‘dog- bone’ tensile specimens were tested to failure. The INSTRON testing machine recorded both the load and the crosshead displacement. When soft elastomeric materials are to be tested, contacting strain measurement techniques, such as strain gauging and ‘clip-on’ mechanical extensometers, are not recommended in general. The reason is that their weight and/or method of attachment can influence the results and the point of failure. Additionally, most mechanical extensometers have limited travel and require removing from the specimen before the fracture occurs. However, the non-contacting strain measurement techniques offer the great advantage of measuring the actual strain on the 6
gauge length of the specimen, when using flexible materials, without any interaction over a very large strain range. Thus, a digital camera monitoring the separation of the two lines inscribed on the test specimen (see Fig. 1) which defines the gauge length, was used. The digital camera was set to take pictures of the gauge length every 10 s recording the change in separation of the two lines throughout the test.
Fig. 1 : Bulk adhesive specimen under testing The digital images were then analyzed using an image processing and analysis software and the engineering strain to failure was extracted for each specimen. The results issued from this test are presented in table 2 [15].
Aluminum-copper alloy ≥ 190 Yield strength (MPa) 17 Elongation at yield stress (%) 0.33 Poisson’s ratio 26500 Shear modulus (MPa)
7
Table 1. Properties
Young’s modulus (MPa)
68000
Tensile of
Aluminium [15].
Table 2. Adhesive properties [15].
Temperature (°C)
Young’s modulus (MPa)
Tensile strength (MPa)
Tensile strain (%)
Room Temperature
2.15 ± 0.8
5.40 ± 0.15
230 ± 0.16
2.2. Specimen preparation In order to make a comparison between the static and fatigue durability of the aluminium alloy bonded assemblies with a one-component polyurethane adhesive, the same design parameters were adopted as previously [15]. Hence, four different state of surface roughness were prepared on aluminium specimens, with an arithmetic average height from Ra ≈ 0.6 µm to Ra ≈ 1.5 µm. Four adhesive thicknesses (e) ranging from 0.3 mm to 2 mm were manufactured.
SLJs specimens were used for fatigue-cyclic tests. Thus, the same geometry, protocol for the preparation of the surface and the manufacture of the specimens as that of the monotonic tests were used. Finally, the metrological conditions are identical to those used in static tests [15].
8
The lap joint geometry used in this study is shown schematically in Fig.2. The width of the joint was 25mm and the aluminium sheet was 3 mm thick.
Fig. 2. Single lap joint geometry (dimensions in mm).
2.3. Fatigue testing Experimental testing was performed in an servo-hydraulic machine under constant load control. The maximum fatigue load was set to 60% of the highest static failure load (Pmax ≈ 1500N), the frequency was 3.5 Hz and an R-ratio of 0.1 was imposed (where R = P*min/P*max and P*min and P*max are the minimum and maximum loads applied over the fatigue cycle, respectively) (Table 3). At least four specimens were tested for each configuration of different combinations of adhesive thickness (0.3, 0.5, 1 and 2 mm) and surface condition (0.6; 1; 1.2 and 1.5 μm). In each test, the number of cycles until failure was recorded. All tests were carried out at an ambient temperature of 25° C and a relative humidity of 40% (r.h.). The criterion of the end of life chosen is the fracture of the bonded joint. Table 3. Maximum load and corresponding shear strength of the loading rate.
% of ultimate load
R-ratio
Maximum load P*max (N)
60
0.1
900
2.4. Scanning electron microscopy analysis (SEM)
9
Maximum shear strength τ*max (MPa) 2.7
Scanning electron microscope (SEM) analyses were performed in a JOEL JSM 6301F/Oxford INCA Energy 350/Gatan Alto 2500 microscope (Tokyo, Japan) at CEMUP (University of Porto, Portugal). This equipment was used to analyse the surface fractures. Samples were coated with an Au/Pd thin film, by sputtering, using the SPI Module Sputter Coater equipment, for 120 sec and with a 15mA current.
3. Experimental results and discussion 3.1. Analysis of the lifetime of bonded assemblies Fig. 3 shows the results of fatigue resistance from SLJs as a function of adhesive thickness and surface roughness. Hence, the results obtained for the different combinations reflect the complication of such a test. Moreover, each couple studied (surface roughness and adhesive thickness) show an important scatter.
10
Fig. 3. Number of cycles to failure as a function of adhesive thickness and surface roughness.
It can be observed, first of all, that with a roughness of Ra ≈ 0.6 μm a maximum lifetime was obtained with an adhesive thickness of 1 mm. The strength of this assembly then decreased by 80% between the thicknesses 1 and 2 mm. For specimens with a roughness Ra ≈ 1 μm, it is clear that the lifetime of the samples increases by increasing the thickness from 0.3 to 0.5 mm. Then, the number of cycles up to failure is almost constant (with a slight decrease of 0.15%) between e = 0.5 mm and e = 1 mm. Unfortunately, the lifetime decreases with a bondline of e = 2 mm. For the tests carried out with a surface roughness Ra ≈ 1.2 μm, the maximum resistance obtained was with an adhesive thickness e = 0.5 mm. Further, the results obtained with bonded specimens which have a surface roughness Ra ≈ 1.5 μm give maximum resistance with a joint thickness of 0.3 mm, which subsequently decreases with 11
thicknesses of 0.5 mm. Then, the strength increases again for e = 1 mm and remains practically constant by increasing the thickness of the bondline to 2 mm. Indeed, these last observations are in agreement with the results of Azari et al. [16] who evaluated the quasi-static and cyclic behaviour of a ductile epoxy adhesive as a function of the adhesive thickness of bonded aluminium assembly in mode-I and mixed mode tests.
Furthermore, a large scatter can be noticed in the number of cycles to failure which does not seem to originate from the bonded assembly as the fractures are all identically mixed (interfacial/cohesive) for the different roughnesses studied with the two adhesive thicknesses e = 0.3 mm and e = 0.5 mm. However, fatigue tests are known for presenting a high scatter [12].
The failure surfaces (mixed failure) of SLJs with 0.3 mm and 0.5 mm thickness for the surface roughness 1.2 and 1.5 µm are shown in Fig. 4a and Fig. 4b respectively. Specimens with a roughness of 0.6 μm and an adhesive thickness e = 0.5 mm show cohesive failure. Likewise, for specimens with e = 1 mm and e = 2 mm the fracture was cohesive for the various roughness tested (Fig. 4c).
12
Fig. 4. Failure mode as a function of adhesive thickness and surface roughness for the cyclic fatigue tests (a) SLJ with 0.3 mm adhesive thickness and 1.5 µm surface roughness, (b) SLJ with 0.5 mm adhesive thickness and 1.2 µm surface roughness, (c) typical cohesive failure.
To better analyse this phenomenon, the study was limited to those results that give the lowest fluctuations. Thus, two roughness was retained, namely: Ra ≈ 0.6 μm and Ra ≈ 1 μm. Fig.5 illustrates the number of cycles to failure of the SLJs as a function of those roughnesses with adhesive thicknesses of e = 0.3 mm, e = 0.5 mm, e = 1 mm and e = 2 mm. Focusing the analysis on these combinations allow to identify the effects of the design parameters of the bonded joint on its lifetime.
13
Fig. 5. Variation of the durability of bonded joints under cyclic fatigue load as a function of (a) adhesive thickness 0.3; 0.5; 1 and 2 mm (b) surface roughness 0.6 and 1 μm.
From the map presented in Fig. 5, useful information can be deduced. The first is associated with the fracture mode since the cohesive and mixed failure are related to the thickness of the joint and the surface roughness (Fig. 5a). The results of the fracture surfaces with a mixed failure are in agreement with the investigation of Azari et al. [17]. 14
When the crack is very close to the interface, an increase in the surface roughness causes deflection of the path of the crack. Hence, a bifurcation of the propagation of the crack within the joint is evident. Thereby, there is an increase of the area of fracture, which decreases the strength of the bonded assembly.
Furthermore, an increase in surface roughness reduces the wettability of the surfaces and, consequently, the strength of the joints [18]. Indeed, it was noticed that for the specimens with a surface roughness of 0.6 μm and an adhesive thickness of 0.5 mm, a cohesive failure occurs. Those results prove the good wettability of the substrates [15]. Otherwise, for the specimens with a surface roughness Ra ≈1 μm, a proportional decrease in the strength of the joints is noticed with the different adhesive thicknesses analysed (Fig. 5b).
The second type of information, to be extracted from these results, is associated with the thickness of the adhesive. In fact, an optimum fatigue life cycle was with specimens with Ra ≈ 0.6 μm, and an adhesive thickness e = 1 mm. These observations are in agreement with the findings of Abou-Hamda et al. [7]. They have proved that the larger fatigue crack growth resistance was obtained with an adhesive thickness e = 1 mm.
On the other hand, in Fig. 5a, a reduction in the number of cycles up to failure was registered with an adhesive thickness of 2 mm. This result can be explained by an increase of the stress at the interface as the adhesive thickness gets thicker [15], which facilitates the possible propagation of the crack. Besides, the loss of the mechanical properties of a thick joint (more than 1 mm) is usually explained by the greater probability of finding a critical size defect. These interpretations are in harmony with the investigations of Kwon et al. [19]. 15
3.2. Statistical investigation of the parameters combination In order to study the effects of the interaction between the two parameters used in this paper, namely the adhesive thickness and the surface roughness, and their influence on the durability of the bonded aluminium assembly under cyclic fatigue load, a statistical investigation of the parameters combination was made. MATLAB® software was used to plot the 3D response surface as shown in Fig 6. As can be seen, the maximum lifetime (7574 cycles) was obtained for an adhesive thickness of 1 mm coupled with the minimum surface roughness of 0.6 μm.
Fig. 6. Influence of the interaction between the adhesive thickness (mm) and the surface roughness (μm) over the lifetime of a bonded aluminium assembly under cyclic fatigue load.
On the other hand, Fig. 7 shows that regardless of the surface roughness used, the lifetime increases with the increase of thickness e = 0.3 mm to e = 1 mm. Then, it decreases when the adhesive thickness increases to 2 mm. In addition, the greatest fatigue life is obtained with a surface roughness Ra ≈ 0.6 μm, regardless of the thickness of the adhesive used.
16
With the increase in surface roughness up to Ra ≈ 1 μm, the strength of the bonded aluminium assembly decreases. These results are in agreement with previous research [20–23], and show that the lifetime of bonded assemblies is highly dependent on the parameters of the bonded surfaces.
Fig. 7. Main effect curves for the average of the number of cycles until failure using the "Mean data".
3.3. Results of the SEM analysis of fracture surfaces SLJs fracture surfaces have either a cohesive failure or mixed (interfacial/cohesive) one. From a macroscopic trend, at the end of the test, sudden ruptures of test specimens were observed, regardless of the combination (roughness/thickness of the adhesive layer) studied. From a microscopic point of view, the failure mechanism that governs damage in the adhesive layer is not known. To compensate for this deficiency, a microscopic analysis by SEM was carried out on fracture surfaces subjected to cyclic fatigue under the same loading conditions at different scales of observations.
17
Cohesive failure (e = 1 mm) Fig. (8a -8d) illustrate the microscopic results of fracture surfaces obtained with surface roughnesses of Ra ≈ 0.6 μm, Ra ≈ 1 μm, Ra ≈ 1.2 μm and Ra ≈ 1.5 μm respectively and an adhesive thickness e = 1 mm. The first observation is the presence of porosities (air bubbles) on the fracture surfaces as well as areas with ligaments caused by the microcracking of the adhesive. The fracture mode, in fatigue, for these four roughnesses and for the same adhesive thickness are very similar. Moreover, it appears to be governed by the same mechanism.
Fig. 8. SEM fracture surfaces for the same adhesive thickness e = 1 mm and surface roughness (a) Ra ≈ 0.6 μm (b) Ra ≈ 1 μm (c) Ra ≈ 1, 2 μm (d) Ra ≈ 1.5 μm.
Fig. 9 shows the formation of cracks and ligaments through different magnification of a fracture surface for Ra ≈ 0.6 μm and e = 1 mm. The latter indicates crack growth phases during the fatigue cycle.
18
Fig. 9. Cohesive fracture surface for test specimens with Ra ≈ 0.6μm and joint thickness e = 1mm (a) fracture surface, (b), (c) and (d) different magnification of SEM fractography.
The fractography of a fracture surface of a specimen with Ra ≈ 1 μm and e = 1 mm are given in Fig.10. Besides, in Fig. 10a, fractography reveals numerous ligaments on the whole of the surface (marked by the arrows on the image). In addition, there is fractures by debonding of particles (marked by circles) with the generation of ligaments due to cracking. Furthermore, the generation of a sharper crack in the vicinity of these detached particles can also be noted, creating fracture initiators. The porosities have areas of crack initiation. Additionally, for the adhesive thickness (e ≥ 1 mm) and for the various 19
roughness tested, the porosities introduce a decrease in the useful section of the sample and locally increase the effective stress applied in the adhesive. This particular phenomenon, driven by the presence of numerous porosities, is perfectly illustrated in Fig. 10b. This fractography shows a clusters of porosities forming clusters facilitating the growth of the crack. These results are in line with the observations of Shenoy et al.[24].
Fig. 10. Cohesive fracture for test specimen with Ra ≈1μm and joint thickness e = 1mm obtained with SEM fractography (a) 2mm magnification, arrows indicate the presence of ligaments and circles the presence of particles (b) magnifications 200 µm.
20
These microscopic observations, confirm the hypotheses previously mentioned. We can conclude that for an adhesive thickness e ≥ 1 mm the propagation of the crack is governed by the presence of a large number of pores (air bubbles) and micro-cracks within the adhesive joint. Mixed (interfacial / cohesive) failure (e = 0.3 mm) Fig.11 shows a fractography of the fracture surfaces for specimens with an adhesive thickness e = 0.3 mm for the four roughnesses tested. This image reveals that the pore size is larger than that observed previously. Regarding the fracture scenario, it can be affirmed that the presence of numerous porosities in one zone causes the increase of the stress locally. The SEM fracture surfaces and the observation of specimens’ failure during the tests can explain the failure path. Damage initially takes place at the substrate/adhesive interface, and then, the adhesive (i.e. in the zone having a cluster of porosities (Fig. 11(b-d)). In this case, the porosity rate is too low to control the cracking paths. Therefore, it is the stress concentrations caused by the roughness of the surfaces at the interface that govern the crack’s growth.
The porosity is mainly generated by the crosslinking process and the low wettability with adhesive thickness e = 0.3 mm. The porosities represent ~ 20% of the area observed. Fig. 11shows porosities of different diameters.
21
Fig. 11. SEM fracture surface for an adhesive thickness e = 0.3 mm (a) Ra ≈ 0,6 .µm (b) Ra ≈ 1 .µm (c) Ra ≈ 1.2 μm (d) Ra ≈ 1.5 μm.
The high viscosity of the one-component polyurethane adhesive and the presence of the roughness of the surfaces prevent the melting of the pores between them, generating twin bubbles rather than a single bubble of larger size. In this case, these porosities do not have any zones showing initiation and propagation of a crack and probably did not initiate damage of the adhesive. On the whole of the surface, the presence of numerous bubbles at the level of the fracture surfaces clearly indicates the weakening of the adhesive layer, but do not seem to be at the origin of the rupture (no ligaments in the porous neighbourhoods). This phenomenon was also evoked by Gupta et al. [25] and 22
Shenoy et al. [26] in the case of microscopic study of the fatigue failure of bonded assemblies using an epoxy. EDS analysis of fracture surfaces An energy dispersive X-ray spectrometry (EDS) analysis was carried out on the fracture surfaces in order to validate the mechanisms of the ruptures presented above. The technical documentation indicates that the particles present in the adhesive are mainly carbon (C). To corroborate this information, an EDS analysis was carried out on a particle of the observed rupture surface. The result of the analysis indicates the presence of the elements C (carbon), O (oxygen), Al (aluminium), Si (silicon), S (sulfur), and Ca (calcium) on the different surfaces of the fractures analysed. Fig. 12 illustrates an analysis performed on a fracture surface of a specimen with Ra ≈ 0.6 μm and e = 1 mm. Observations on two spectrum prove that the particles present are derived from the polyurethane adhesive corresponding to the high peaks of the elements C and O.
23
Fig. 12. Specimen with Ra≈0.6 μm and e=1 mm. X Analysis representing the number of shots carried by the electrons on the surface of the particle as a function of the energy in Kev. (the energy level determines the element in the presence).
Fig. 13 shows the EDs analysis for a mixed failure surfaces Ra ≈ 0.6 μm and e = 0.3 mm. A significant peak of element C is observed on "spectrum 1" indicating that the particle is largely composed of carbon, as well as a lower peak of element O and Ca. This is the polyurethane adhesive. However, on "spectrum 2", a peak of element C is also observed as well as a smaller peak of the element O, Al, S and Si. The presence of the element Si may originate from the silane containing the Si element in its composition. The adhesive applied to the silaned surface (primer adhesion) may have been mixed with the silane and thus revealed the presence of the Si element. The presence of the element Al allows us to confirm that this is indeed the surface of the substrate. This confirms the mixed failure (interfacial/cohesive).
24
Fig. 13. Specimen with Ra≈ 0.6 μm and e = 0.3 mm. X Analysis represents the number of shots carried by the electrons on the surface of the particle as a function of the energy in keV.
4. Conclusion In this paper the performance of a one-component polyurethane adhesive, used in the automotive industry for aluminium bonding, under cyclic fatigue tests was investigated. The effect of surface roughness and adhesive thickness on the lifetime of the bonded assembly was analysed. Fatigue results showed that the durability of the joint was strongly affected by both the surface roughness and the joint thickness. Moreover, it was found that damage observed after fatigue failure is very similar to that obtained for monotonic testing. In fact, there is an optimum surface roughness Ra ≈ 0.6 µm and adhesive thickness e = 1 mm allow the highest lifetime of bonded aluminium assembly.
25
Acknowledgements These research and innovation are made in the context of a MOBIDOC thesis financed by the EU within the framework of the PASRI program. The authors would like to thank ICAR industry Tunisia, for supporting the work here presented.
References [1]
Kinloch AJ. Adhesives in engineering. Proc Inst Mech Eng Part G 1997;211:307– 35.
[2]
Mazumdar SK, Mallick K. Static and Fatigue Behavior of Adhesive Joints in SMC-SMC Composites. Polym Compos 1998;19:139–46.
[3]
Curley AJ, Hadavinia H, Kinloch AJ, Taylor AC. Predicting the service-life of adhesively-bonded
joints.
Int
J
Fract
2000;103:41–69.
doi:10.1023/A:1007669219149. [4]
Rushforth MW, Bowen P, McAlpine E, Zhou X, Thompson GE. The effect of surface pretreatment and moisture on the fatigue performance of adhesivelybonded
aluminium.
J
Mater
Process
Technol
2004;153–154:359–65.
doi:10.1016/j.jmatprotec.2004.04.319. [5]
Abel ML, Adams ANN, Kinloch AJ, Shaw SJ, Watts JF. The effects of surface pretreatment on the cyclic-fatigue characteristics of bonded aluminium-alloy joints. Int J Adhes Adhes 2006;26:50–61. doi:10.1016/j.ijadhadh.2004.12.004.
[6]
Da Silva LFM, Ferreira NMAJ, Richter-Trummer V, Marques EAS. Effect of grooves on the strength of adhesively bonded joints. Int J Adhes Adhes 2010;30:735–43. doi:10.1016/j.ijadhadh.2010.07.005.
[7]
Abou-Hamda MM, Megahed MM, Hammouda MMI. Fatigue crack growth in double cantilever beam specimen with an adhesive layer. Eng Fract Mech 1998;60:605–14. doi:10.1016/S0013-7944(98)00018-6.
[8]
Banea MD, da Silva LFM. Static and fatigue behaviour of room temperature vulcanising silicone adhesives for high temperature aerospace applications. Statisches
Verhalten
und
Dauerfestigkeitsanalyse
von
vulkanisierten
Silikonklebstoffen für Luftfahrtanwendungen bei hohen Temperat. Materwiss 26
Werksttech 2010;41:325–35. doi:10.1002/mawe.201000605. [9]
Abdel Wahab MM. Fatigue in Adhesively Bonded Joints: A Review. ISRN Mater Sci 2012;2012:1–25. doi:10.5402/2012/746308.
[10]
Costa M, Viana G, da Silva LFM, Campilho RDSG. Environmental effect on the fatigue degradation of adhesive joints: A review. J Adhes 2017;93:127–46. doi:10.1080/00218464.2016.1179117.
[11]
Costa M, Viana G, da Silva L, Campilho R. Effect of humidity on the fatigue behaviour of adhesively bonded aluminium joints. Lat Am J Solids Struct 2017;14:174–87. doi:10.1177/1464420716645263.
[12]
Barros S De, Kenedi PP, Ferreira SM, Budhe S, Bernardino AJ. Influence of mechanical surface treatment on fatigue life of bonded joints. J Adhes 2017;93:599–612. doi:10.1080/00218464.2015.1122531.
[13]
Loureiro a. L, da Silva LFM, Sato C, Figueiredo M a. V. Comparison of the Mechanical Behaviour Between Stiff and Flexible Adhesive Joints for the Automotive
Industry.
J
Adhes
2010;86:765–87.
doi:10.1080/00218464.2010.482440. [14]
Boutar Y, Naïmi S, Mezlini S, da Silva LFM, Ben Sik Ali M. Characterization of aluminium one-component polyurethane adhesive joints as a function of bond thickness for the automotive industry: Fracture analysis and behavior. Eng Fract Mech 2017;177:45–60. doi:10.1016/j.engfracmech.2017.03.044.
[15]
Boutar Y, Naïmi S, Mezlini S, da Silva LFM, Hamdaoui M, Ben Sik Ali M. Effect of adhesive thickness and surface roughness on the shear strength of aluminium one-component
polyurethane
applications.
Int
J
adhesive Adhes
single-lap Sci
joints
Technol
for
automotive
2016;4243:1–17.
doi:10.1080/01694243.2016.1170588. [16]
Azari S, Papini M, Spelt JK. Effect of adhesive thickness on fatigue and fracture of toughened epoxy joints - Part I: Experiments. Eng Fract Mech 2011;78:153–62. doi:10.1016/j.engfracmech.2010.06.025.
[17]
Azari S, Papini M, Spelt JK. Effect of Surface Roughness on the Performance of Adhesive Joints Under Static and Cyclic Loading. J Adhes 2010;86:742–64. doi:10.1080/00218464.2010.482430. 27
[18]
Boutar Y, Naïmi S, Mezlini S, Ali MBS. Effect of surface treatment on the shear strength of aluminium adhesive single-lap joints for automotive applications. Int J Adhes Adhes 2016;67:38–43. doi:10.1016/j.ijadhadh.2015.12.023.
[19]
Kwon JW, Lee DG. The effects of surface roughness and bond thickness on the fatigue life of adhesively bonded tubular single lap joints. J Adhes Sci Technol 2000;14:1085–102. doi:10.1163/156856100743095.
[20]
Blanchard C, Chateauminois A, Vincent L. A new testing methodology for the assessment of fatigue properties of structural adhesives. Int J Adhes Adhes 1996;16:289–99. doi:10.1016/S0143-7496(96)00018-8.
[21]
Pereira AM, Ferreira JM, Antunes F V., Bártolo PJ. Study on the fatigue strength of AA 6082-T6 adhesive lap joints. Int J Adhes Adhes 2009;29:633–8. doi:10.1016/j.ijadhadh.2009.02.009.
[22]
Underhill PR, Rider AN, Duquesnay DL. The effect of warm water surface treatments on the fatigue life in shear of aluminum joints. Int J Adhes Adhes 2006;26:199–205. doi:10.1016/j.ijadhadh.2004.10.005.
[23]
Pascoe JA, Alderliesten RC, Benedictus R. Methods for the prediction of fatigue delamination growth in composites and adhesive bonds – A critical review. Eng Fract Mech 2013;112–113:72–96. doi:10.1016/j.engfracmech.2013.10.003.
[24]
Shenoy V, Ashcroft IA, Critchlow GW, Crocombe AD, Abdel Wahab MM. Strength wearout of adhesively bonded joints under constant amplitude fatigue. Int J Fatigue 2009;31:820–30. doi:10.1016/j.ijfatigue.2008.11.007.
[25]
Gupta VB, Drzal LT, Adams WW, Omlor R. An electron microscopic study of the morphology
of
cured
epoxy
resin.
J
Mater
Sci
1985;20:3439–52.
doi:10.1007/BF01113751. [26]
Shenoy V, Ashcroft IA, Critchlow GW, Crocombe AD, Wahab MMA. An investigation into the crack initiation and propagation behaviour of bonded singlelap joints using backface strain. Int J Adhes Adhes 2009;29:361–71. doi:10.1016/j.ijadhadh.2008.07.008.
28
PII: DOI: Reference:
S0143-7496(18)30045-9 https://doi.org/10.1016/j.ijadhadh.2018.02.012 JAAD2137
To appear in: International Journal of Adhesion and Adhesives Cite this article as: Y. Boutar, S. Naïmi, S. Mezlini, R.J.C. Carbas, L.F.M. da Silva and M. Ben Sik Ali, Fatigue resistance of an aluminium one-component polyurethane adhesive joint for the automotive industry: Effect of surface roughness and adhesive thickness, International Journal of Adhesion and Adhesives, https://doi.org/10.1016/j.ijadhadh.2018.02.012 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Fatigue resistance of an aluminium one-component polyurethane adhesive joint for the automotive industry: Effect of surface roughness and adhesive thickness Y. Boutar 1*, S. Naïmi 1, S. Mezlini 1, R.J.C. Carbas2, L.F.M. da Silva2, M. Ben Sik Ali3 1
Mechanical Engineering Laboratory, National Engineering School of Monastir, Monastir University, Monastir, Tunisia 2
Faculdade de Engenharia, Departamento de Engenharia Mecânica, Universidade do Porto, Porto, Portugal 3
*
Automotive Industry, ICAR, Sousse, Tunisia
Corresponding author, Email:[email protected]
Abstract Fatigue is a crucial type of loading for many structural components that contain adhesive bonding systems such as the automotive industry. One of the main advantages of adhesives is their weight-reduction aptitudes. Furthermore, adhesively bonded joints allow a good damping to the fatigue's solicitation with fewer sources of stress concentration. In addition, a suitable surface treatment allows a great joint interface adhesion under cyclic loading. Hence, the mechanical preparation of the bonded surface influences directly the bonded joint performance. Therefore, a careful consideration should be given to the choice of the bonded adhesive thickness. The aim of this paper is to investigate the influence of surface roughness of an aluminium alloy and the adhesive thickness of a one-component polyurethane adhesive, to be used in several parts for buses structure, on the fatigue behaviour of single lap joints. To achieve this purpose, four different surface roughnesses were prepared on aluminium specimens, with an arithmetic average height from Ra ≈ 0.6 µm to Ra ≈ 1.5 µm. Bonded specimens with four adhesive 1
thicknesses from 0.3 mm to 2 mm were manufactured and tested in dynamic tests. The results confirm that there is an optimum combination of surface roughness and adhesive thickness providing the maximum fatigue life. Moreover, good agreement was found between this investigation and a previous work using the same parameters under static loading.
Keywords Single lap joint; Polyurethane adhesive; Surface roughness; Adhesive thickness; Fatigue life.
1. Introduction The use of adhesive bonding in the industry has greatly increased in recent years. However, its use in truly structural applications is still often limited. This is mainly due to a lack of confidence in the performance of adhesive joints, since the mechanical performance of the joints may deteriorate upon being subjected to cyclic-fatigue loading [1]. Therefore, an understanding of the effect of substrate's surface roughness and bondline thickness on the cyclic loading behaviour of adhesive joints can help to select an accurate surface treatment and an optimal adhesive thickness. Thus, the ability to quantitatively describe the reduction in performance and to predict the lifetime of bonded joints would be a powerful tool. This enables manufacturers to make wider and more efficient use of adhesive bonding.
In that context, there has been many papers dealing with the effect of bondline thickness and surface treatments under cyclic loading. Mazumdar and Mallick [2] discuss the static
2
and fatigue behaviour of adhesively bonded single lap joints in Sheet Molding Compound SMC-SMC composites. By studying the effects of lap length and adhesive thickness, the authors show that these two parameters have a negligible effect on the ratio of fatigue strength to static strength. The fatigue strength at 106 cycles is approximately 50% to 54% of the static strength for various adhesive thicknesses and lap lengths investigated in this study.
Moreover, in order to predict the cyclic-fatigue performance of single-lap joints and a typical bonded component, a fracture-mechanics approach was used by Curley et al. [3]. In their work, mild-steel substrates and a rubber-toughened hot-curing epoxy adhesive were used. For lap joints subjected to the ‘wet’ fatigue tests, the joints visually exhibited an apparent interfacial failure. But, for the tests undertaken in the ‘dry’ environment, failure occurred mainly in the adhesive layer, but very close to the interface. The researchers prove that the results from the fracture-mechanics tests not only provided the experimental data needed for the modelling studies but also indicated the expected failure path for the lap joints, which greatly assisted in the detailed modelling of the location of the cracks in these joints. Rushforth et al. [4] investigated the effect of surface pretreatment and moisture on the fatigue performance of adhesively-bonded aluminium. Single lap joints (SLJs) were tested to failure under fatigue loading to ascertain the effects of surface pre-treatment and moisture in both durability and fracture path. The aluminium substrate used was AA5754 with and without a silicon-based pre-treatment. The fatigue performance of pretreated and untreated joints was similar when tested in air, but the performance of the untreated joints was greatly reduced when tested in 96% relative humidity (r.h.) and temperature at 26±1°C. The differences in fatigue 3
performance between untreated and pretreated joints in 96% (r.h.) are thought to be due to differences in the failure mode of the joints arising from the improved interfacial bonding between the pre-treatment and the adhesive. Pretreated joints failed predominantly cohesively through the adhesive, while untreated joints failed mainly interfacially at the adhesive/aluminium interface. Abel et al. [5] proposed an accelerated test method, based upon cyclic-fatigue testing using a fracture-mechanics approach, to study the durability of organosilane-pretreated joints which were adhesively-bonded using a hot-cured epoxy-film adhesive. The silane primer used as a pre-treatment was shown to be effective in increasing the joint durability, compared with a simple grit-blast and degrease (GBD) treatment and, indeed, gives comparable results to using a chromicacid etch (CAE) pre-treatment. Furthermore, da Silva et al.[6] studied single lap joints with patterns under cyclic loading with and without chemical surface treatment and bonded with a brittle adhesive. Fatigue testing showed that the surface patterns have a beneficial effect on the behaviour of the specimens under cyclic conditions. At the same load, patterned specimens consistently endured more cycles than the specimens with no pattern, when no additional chemical surface treatment was applied.
Likewise, there are several papers that deal with the effect of bondline thickness or both surface treatment and adhesive thickness on the behaviour of cycle fatigue tests for bonded joint. In the paper of Abou-Hamad et al. [7], an experimental method based on fatigue crack growth of an aluminium bonded DCB specimens was adopted. A twocomponent epoxy was used. Three bondline thickness was investigated, namely: 0.3, 0.8 and 1 mm. The results show that the increase of adhesive thickness has a significant effect on fatigue crack growth. The larger the bond line thickness, the larger is the fatigue 4
crack growth resistance. Banea et al. [8], examined the mechanical properties of room temperature vulcanising (RTV) silicone adhesives. SLJs were fabricated and tested to assess the adhesive performance in a joint. The influence of adhesive thickness (0.5 and 1 mm) on the fatigue behaviour was made. The fatigue tests on RTV silicone adhesives SLJs showed a fatigue limit of approximately 30% of the static failure load. Moreover, Abdel Wahab [9] and Costa et al. [10,11], showed that one of the most parameters that affect fatigue strength and lifetime of bonded assembly is the surface treatment of the adherends. Therefore, these parameters need to be optimised to guarantee the bonded assembly durability. Recently, de Barros et al. [12] analysed the changes in fatigue resistance due to surface preparation in terms of a total number of cycles, further confirming that the adherends surfaces need to be carefully prepared prior to bonding.
However, modern polyurethane adhesives are increasingly being used in structural joining applications as they present important advantages in terms of damping, impact, fatigue, and safety which are critical factors in the automotive industry [13]. On the other hand, most of the results from the literature regarding the effect of surface treatment and adhesive thickness are for typical structural epoxy adhesives, which are generally formulated to perform in thin sections, while polyurethane adhesives are designed to perform in thicker sections and might have a different behaviour as a function of the design parameters. Therefore, it is important to know how polyurethane adhesive-bonded joints performance is affected by the geometric parameters. A recent static study by Boutar et al. [14,15], on the effect of surface roughness and adhesive thickness for an aluminium one-component polyurethane bonded assembly, show that there is an optimum couple of parameter allowing the highest shear strength for this joints. Hence, it 5
is worthy to distinguish how polyurethane adhesive-bonded joints performance is affected by surface roughness and adhesive thickness when they are subjected to cyclic fatigue loading.
In this study, four different surface roughness were prepared on aluminium specimens, with an arithmetic average height from Ra ≈ 0.6 µm to Ra ≈ 1.5 µm. Bonded SLJs specimens with four adhesive thicknesses from 0.3 mm to 2 mm were manufactured and tested in dynamic tests. During tests, the number of life cycles was recorded. Fracture surfaces were examined visually and using high-powered microscope SEM and EDS so that a detailed picture of the failure mechanism could be obtained.
2. Experimental details 2.1. Material The materials used in this study were an aluminium–copper alloy 5754 (see Table 1) and a one-component polyurethane adhesive DINITROL 500. Indeed, The Dinitrol 500 adhesive specimens were tested at Room Temperature in tension using a testing machine INSTRON model 3367, under a constant crosshead rate of 5mm/min. Three ‘dog- bone’ tensile specimens were tested to failure. The INSTRON testing machine recorded both the load and the crosshead displacement. When soft elastomeric materials are to be tested, contacting strain measurement techniques, such as strain gauging and ‘clip-on’ mechanical extensometers, are not recommended in general. The reason is that their weight and/or method of attachment can influence the results and the point of failure. Additionally, most mechanical extensometers have limited travel and require removing from the specimen before the fracture occurs. However, the non-contacting strain measurement techniques offer the great advantage of measuring the actual strain on the 6
gauge length of the specimen, when using flexible materials, without any interaction over a very large strain range. Thus, a digital camera monitoring the separation of the two lines inscribed on the test specimen (see Fig. 1) which defines the gauge length, was used. The digital camera was set to take pictures of the gauge length every 10 s recording the change in separation of the two lines throughout the test.
Fig. 1 : Bulk adhesive specimen under testing The digital images were then analyzed using an image processing and analysis software and the engineering strain to failure was extracted for each specimen. The results issued from this test are presented in table 2 [15].
Aluminum-copper alloy ≥ 190 Yield strength (MPa) 17 Elongation at yield stress (%) 0.33 Poisson’s ratio 26500 Shear modulus (MPa)
7
Table 1. Properties
Young’s modulus (MPa)
68000
Tensile of
Aluminium [15].
Table 2. Adhesive properties [15].
Temperature (°C)
Young’s modulus (MPa)
Tensile strength (MPa)
Tensile strain (%)
Room Temperature
2.15 ± 0.8
5.40 ± 0.15
230 ± 0.16
2.2. Specimen preparation In order to make a comparison between the static and fatigue durability of the aluminium alloy bonded assemblies with a one-component polyurethane adhesive, the same design parameters were adopted as previously [15]. Hence, four different state of surface roughness were prepared on aluminium specimens, with an arithmetic average height from Ra ≈ 0.6 µm to Ra ≈ 1.5 µm. Four adhesive thicknesses (e) ranging from 0.3 mm to 2 mm were manufactured.
SLJs specimens were used for fatigue-cyclic tests. Thus, the same geometry, protocol for the preparation of the surface and the manufacture of the specimens as that of the monotonic tests were used. Finally, the metrological conditions are identical to those used in static tests [15].
8
The lap joint geometry used in this study is shown schematically in Fig.2. The width of the joint was 25mm and the aluminium sheet was 3 mm thick.
Fig. 2. Single lap joint geometry (dimensions in mm).
2.3. Fatigue testing Experimental testing was performed in an servo-hydraulic machine under constant load control. The maximum fatigue load was set to 60% of the highest static failure load (Pmax ≈ 1500N), the frequency was 3.5 Hz and an R-ratio of 0.1 was imposed (where R = P*min/P*max and P*min and P*max are the minimum and maximum loads applied over the fatigue cycle, respectively) (Table 3). At least four specimens were tested for each configuration of different combinations of adhesive thickness (0.3, 0.5, 1 and 2 mm) and surface condition (0.6; 1; 1.2 and 1.5 μm). In each test, the number of cycles until failure was recorded. All tests were carried out at an ambient temperature of 25° C and a relative humidity of 40% (r.h.). The criterion of the end of life chosen is the fracture of the bonded joint. Table 3. Maximum load and corresponding shear strength of the loading rate.
% of ultimate load
R-ratio
Maximum load P*max (N)
60
0.1
900
2.4. Scanning electron microscopy analysis (SEM)
9
Maximum shear strength τ*max (MPa) 2.7
Scanning electron microscope (SEM) analyses were performed in a JOEL JSM 6301F/Oxford INCA Energy 350/Gatan Alto 2500 microscope (Tokyo, Japan) at CEMUP (University of Porto, Portugal). This equipment was used to analyse the surface fractures. Samples were coated with an Au/Pd thin film, by sputtering, using the SPI Module Sputter Coater equipment, for 120 sec and with a 15mA current.
3. Experimental results and discussion 3.1. Analysis of the lifetime of bonded assemblies Fig. 3 shows the results of fatigue resistance from SLJs as a function of adhesive thickness and surface roughness. Hence, the results obtained for the different combinations reflect the complication of such a test. Moreover, each couple studied (surface roughness and adhesive thickness) show an important scatter.
10
Fig. 3. Number of cycles to failure as a function of adhesive thickness and surface roughness.
It can be observed, first of all, that with a roughness of Ra ≈ 0.6 μm a maximum lifetime was obtained with an adhesive thickness of 1 mm. The strength of this assembly then decreased by 80% between the thicknesses 1 and 2 mm. For specimens with a roughness Ra ≈ 1 μm, it is clear that the lifetime of the samples increases by increasing the thickness from 0.3 to 0.5 mm. Then, the number of cycles up to failure is almost constant (with a slight decrease of 0.15%) between e = 0.5 mm and e = 1 mm. Unfortunately, the lifetime decreases with a bondline of e = 2 mm. For the tests carried out with a surface roughness Ra ≈ 1.2 μm, the maximum resistance obtained was with an adhesive thickness e = 0.5 mm. Further, the results obtained with bonded specimens which have a surface roughness Ra ≈ 1.5 μm give maximum resistance with a joint thickness of 0.3 mm, which subsequently decreases with 11
thicknesses of 0.5 mm. Then, the strength increases again for e = 1 mm and remains practically constant by increasing the thickness of the bondline to 2 mm. Indeed, these last observations are in agreement with the results of Azari et al. [16] who evaluated the quasi-static and cyclic behaviour of a ductile epoxy adhesive as a function of the adhesive thickness of bonded aluminium assembly in mode-I and mixed mode tests.
Furthermore, a large scatter can be noticed in the number of cycles to failure which does not seem to originate from the bonded assembly as the fractures are all identically mixed (interfacial/cohesive) for the different roughnesses studied with the two adhesive thicknesses e = 0.3 mm and e = 0.5 mm. However, fatigue tests are known for presenting a high scatter [12].
The failure surfaces (mixed failure) of SLJs with 0.3 mm and 0.5 mm thickness for the surface roughness 1.2 and 1.5 µm are shown in Fig. 4a and Fig. 4b respectively. Specimens with a roughness of 0.6 μm and an adhesive thickness e = 0.5 mm show cohesive failure. Likewise, for specimens with e = 1 mm and e = 2 mm the fracture was cohesive for the various roughness tested (Fig. 4c).
12
Fig. 4. Failure mode as a function of adhesive thickness and surface roughness for the cyclic fatigue tests (a) SLJ with 0.3 mm adhesive thickness and 1.5 µm surface roughness, (b) SLJ with 0.5 mm adhesive thickness and 1.2 µm surface roughness, (c) typical cohesive failure.
To better analyse this phenomenon, the study was limited to those results that give the lowest fluctuations. Thus, two roughness was retained, namely: Ra ≈ 0.6 μm and Ra ≈ 1 μm. Fig.5 illustrates the number of cycles to failure of the SLJs as a function of those roughnesses with adhesive thicknesses of e = 0.3 mm, e = 0.5 mm, e = 1 mm and e = 2 mm. Focusing the analysis on these combinations allow to identify the effects of the design parameters of the bonded joint on its lifetime.
13
Fig. 5. Variation of the durability of bonded joints under cyclic fatigue load as a function of (a) adhesive thickness 0.3; 0.5; 1 and 2 mm (b) surface roughness 0.6 and 1 μm.
From the map presented in Fig. 5, useful information can be deduced. The first is associated with the fracture mode since the cohesive and mixed failure are related to the thickness of the joint and the surface roughness (Fig. 5a). The results of the fracture surfaces with a mixed failure are in agreement with the investigation of Azari et al. [17]. 14
When the crack is very close to the interface, an increase in the surface roughness causes deflection of the path of the crack. Hence, a bifurcation of the propagation of the crack within the joint is evident. Thereby, there is an increase of the area of fracture, which decreases the strength of the bonded assembly.
Furthermore, an increase in surface roughness reduces the wettability of the surfaces and, consequently, the strength of the joints [18]. Indeed, it was noticed that for the specimens with a surface roughness of 0.6 μm and an adhesive thickness of 0.5 mm, a cohesive failure occurs. Those results prove the good wettability of the substrates [15]. Otherwise, for the specimens with a surface roughness Ra ≈1 μm, a proportional decrease in the strength of the joints is noticed with the different adhesive thicknesses analysed (Fig. 5b).
The second type of information, to be extracted from these results, is associated with the thickness of the adhesive. In fact, an optimum fatigue life cycle was with specimens with Ra ≈ 0.6 μm, and an adhesive thickness e = 1 mm. These observations are in agreement with the findings of Abou-Hamda et al. [7]. They have proved that the larger fatigue crack growth resistance was obtained with an adhesive thickness e = 1 mm.
On the other hand, in Fig. 5a, a reduction in the number of cycles up to failure was registered with an adhesive thickness of 2 mm. This result can be explained by an increase of the stress at the interface as the adhesive thickness gets thicker [15], which facilitates the possible propagation of the crack. Besides, the loss of the mechanical properties of a thick joint (more than 1 mm) is usually explained by the greater probability of finding a critical size defect. These interpretations are in harmony with the investigations of Kwon et al. [19]. 15
3.2. Statistical investigation of the parameters combination In order to study the effects of the interaction between the two parameters used in this paper, namely the adhesive thickness and the surface roughness, and their influence on the durability of the bonded aluminium assembly under cyclic fatigue load, a statistical investigation of the parameters combination was made. MATLAB® software was used to plot the 3D response surface as shown in Fig 6. As can be seen, the maximum lifetime (7574 cycles) was obtained for an adhesive thickness of 1 mm coupled with the minimum surface roughness of 0.6 μm.
Fig. 6. Influence of the interaction between the adhesive thickness (mm) and the surface roughness (μm) over the lifetime of a bonded aluminium assembly under cyclic fatigue load.
On the other hand, Fig. 7 shows that regardless of the surface roughness used, the lifetime increases with the increase of thickness e = 0.3 mm to e = 1 mm. Then, it decreases when the adhesive thickness increases to 2 mm. In addition, the greatest fatigue life is obtained with a surface roughness Ra ≈ 0.6 μm, regardless of the thickness of the adhesive used.
16
With the increase in surface roughness up to Ra ≈ 1 μm, the strength of the bonded aluminium assembly decreases. These results are in agreement with previous research [20–23], and show that the lifetime of bonded assemblies is highly dependent on the parameters of the bonded surfaces.
Fig. 7. Main effect curves for the average of the number of cycles until failure using the "Mean data".
3.3. Results of the SEM analysis of fracture surfaces SLJs fracture surfaces have either a cohesive failure or mixed (interfacial/cohesive) one. From a macroscopic trend, at the end of the test, sudden ruptures of test specimens were observed, regardless of the combination (roughness/thickness of the adhesive layer) studied. From a microscopic point of view, the failure mechanism that governs damage in the adhesive layer is not known. To compensate for this deficiency, a microscopic analysis by SEM was carried out on fracture surfaces subjected to cyclic fatigue under the same loading conditions at different scales of observations.
17
Cohesive failure (e = 1 mm) Fig. (8a -8d) illustrate the microscopic results of fracture surfaces obtained with surface roughnesses of Ra ≈ 0.6 μm, Ra ≈ 1 μm, Ra ≈ 1.2 μm and Ra ≈ 1.5 μm respectively and an adhesive thickness e = 1 mm. The first observation is the presence of porosities (air bubbles) on the fracture surfaces as well as areas with ligaments caused by the microcracking of the adhesive. The fracture mode, in fatigue, for these four roughnesses and for the same adhesive thickness are very similar. Moreover, it appears to be governed by the same mechanism.
Fig. 8. SEM fracture surfaces for the same adhesive thickness e = 1 mm and surface roughness (a) Ra ≈ 0.6 μm (b) Ra ≈ 1 μm (c) Ra ≈ 1, 2 μm (d) Ra ≈ 1.5 μm.
Fig. 9 shows the formation of cracks and ligaments through different magnification of a fracture surface for Ra ≈ 0.6 μm and e = 1 mm. The latter indicates crack growth phases during the fatigue cycle.
18
Fig. 9. Cohesive fracture surface for test specimens with Ra ≈ 0.6μm and joint thickness e = 1mm (a) fracture surface, (b), (c) and (d) different magnification of SEM fractography.
The fractography of a fracture surface of a specimen with Ra ≈ 1 μm and e = 1 mm are given in Fig.10. Besides, in Fig. 10a, fractography reveals numerous ligaments on the whole of the surface (marked by the arrows on the image). In addition, there is fractures by debonding of particles (marked by circles) with the generation of ligaments due to cracking. Furthermore, the generation of a sharper crack in the vicinity of these detached particles can also be noted, creating fracture initiators. The porosities have areas of crack initiation. Additionally, for the adhesive thickness (e ≥ 1 mm) and for the various 19
roughness tested, the porosities introduce a decrease in the useful section of the sample and locally increase the effective stress applied in the adhesive. This particular phenomenon, driven by the presence of numerous porosities, is perfectly illustrated in Fig. 10b. This fractography shows a clusters of porosities forming clusters facilitating the growth of the crack. These results are in line with the observations of Shenoy et al.[24].
Fig. 10. Cohesive fracture for test specimen with Ra ≈1μm and joint thickness e = 1mm obtained with SEM fractography (a) 2mm magnification, arrows indicate the presence of ligaments and circles the presence of particles (b) magnifications 200 µm.
20
These microscopic observations, confirm the hypotheses previously mentioned. We can conclude that for an adhesive thickness e ≥ 1 mm the propagation of the crack is governed by the presence of a large number of pores (air bubbles) and micro-cracks within the adhesive joint. Mixed (interfacial / cohesive) failure (e = 0.3 mm) Fig.11 shows a fractography of the fracture surfaces for specimens with an adhesive thickness e = 0.3 mm for the four roughnesses tested. This image reveals that the pore size is larger than that observed previously. Regarding the fracture scenario, it can be affirmed that the presence of numerous porosities in one zone causes the increase of the stress locally. The SEM fracture surfaces and the observation of specimens’ failure during the tests can explain the failure path. Damage initially takes place at the substrate/adhesive interface, and then, the adhesive (i.e. in the zone having a cluster of porosities (Fig. 11(b-d)). In this case, the porosity rate is too low to control the cracking paths. Therefore, it is the stress concentrations caused by the roughness of the surfaces at the interface that govern the crack’s growth.
The porosity is mainly generated by the crosslinking process and the low wettability with adhesive thickness e = 0.3 mm. The porosities represent ~ 20% of the area observed. Fig. 11shows porosities of different diameters.
21
Fig. 11. SEM fracture surface for an adhesive thickness e = 0.3 mm (a) Ra ≈ 0,6 .µm (b) Ra ≈ 1 .µm (c) Ra ≈ 1.2 μm (d) Ra ≈ 1.5 μm.
The high viscosity of the one-component polyurethane adhesive and the presence of the roughness of the surfaces prevent the melting of the pores between them, generating twin bubbles rather than a single bubble of larger size. In this case, these porosities do not have any zones showing initiation and propagation of a crack and probably did not initiate damage of the adhesive. On the whole of the surface, the presence of numerous bubbles at the level of the fracture surfaces clearly indicates the weakening of the adhesive layer, but do not seem to be at the origin of the rupture (no ligaments in the porous neighbourhoods). This phenomenon was also evoked by Gupta et al. [25] and 22
Shenoy et al. [26] in the case of microscopic study of the fatigue failure of bonded assemblies using an epoxy. EDS analysis of fracture surfaces An energy dispersive X-ray spectrometry (EDS) analysis was carried out on the fracture surfaces in order to validate the mechanisms of the ruptures presented above. The technical documentation indicates that the particles present in the adhesive are mainly carbon (C). To corroborate this information, an EDS analysis was carried out on a particle of the observed rupture surface. The result of the analysis indicates the presence of the elements C (carbon), O (oxygen), Al (aluminium), Si (silicon), S (sulfur), and Ca (calcium) on the different surfaces of the fractures analysed. Fig. 12 illustrates an analysis performed on a fracture surface of a specimen with Ra ≈ 0.6 μm and e = 1 mm. Observations on two spectrum prove that the particles present are derived from the polyurethane adhesive corresponding to the high peaks of the elements C and O.
23
Fig. 12. Specimen with Ra≈0.6 μm and e=1 mm. X Analysis representing the number of shots carried by the electrons on the surface of the particle as a function of the energy in Kev. (the energy level determines the element in the presence).
Fig. 13 shows the EDs analysis for a mixed failure surfaces Ra ≈ 0.6 μm and e = 0.3 mm. A significant peak of element C is observed on "spectrum 1" indicating that the particle is largely composed of carbon, as well as a lower peak of element O and Ca. This is the polyurethane adhesive. However, on "spectrum 2", a peak of element C is also observed as well as a smaller peak of the element O, Al, S and Si. The presence of the element Si may originate from the silane containing the Si element in its composition. The adhesive applied to the silaned surface (primer adhesion) may have been mixed with the silane and thus revealed the presence of the Si element. The presence of the element Al allows us to confirm that this is indeed the surface of the substrate. This confirms the mixed failure (interfacial/cohesive).
24
Fig. 13. Specimen with Ra≈ 0.6 μm and e = 0.3 mm. X Analysis represents the number of shots carried by the electrons on the surface of the particle as a function of the energy in keV.
4. Conclusion In this paper the performance of a one-component polyurethane adhesive, used in the automotive industry for aluminium bonding, under cyclic fatigue tests was investigated. The effect of surface roughness and adhesive thickness on the lifetime of the bonded assembly was analysed. Fatigue results showed that the durability of the joint was strongly affected by both the surface roughness and the joint thickness. Moreover, it was found that damage observed after fatigue failure is very similar to that obtained for monotonic testing. In fact, there is an optimum surface roughness Ra ≈ 0.6 µm and adhesive thickness e = 1 mm allow the highest lifetime of bonded aluminium assembly.
25
Acknowledgements These research and innovation are made in the context of a MOBIDOC thesis financed by the EU within the framework of the PASRI program. The authors would like to thank ICAR industry Tunisia, for supporting the work here presented.
References [1]
Kinloch AJ. Adhesives in engineering. Proc Inst Mech Eng Part G 1997;211:307– 35.
[2]
Mazumdar SK, Mallick K. Static and Fatigue Behavior of Adhesive Joints in SMC-SMC Composites. Polym Compos 1998;19:139–46.
[3]
Curley AJ, Hadavinia H, Kinloch AJ, Taylor AC. Predicting the service-life of adhesively-bonded
joints.
Int
J
Fract
2000;103:41–69.
doi:10.1023/A:1007669219149. [4]
Rushforth MW, Bowen P, McAlpine E, Zhou X, Thompson GE. The effect of surface pretreatment and moisture on the fatigue performance of adhesivelybonded
aluminium.
J
Mater
Process
Technol
2004;153–154:359–65.
doi:10.1016/j.jmatprotec.2004.04.319. [5]
Abel ML, Adams ANN, Kinloch AJ, Shaw SJ, Watts JF. The effects of surface pretreatment on the cyclic-fatigue characteristics of bonded aluminium-alloy joints. Int J Adhes Adhes 2006;26:50–61. doi:10.1016/j.ijadhadh.2004.12.004.
[6]
Da Silva LFM, Ferreira NMAJ, Richter-Trummer V, Marques EAS. Effect of grooves on the strength of adhesively bonded joints. Int J Adhes Adhes 2010;30:735–43. doi:10.1016/j.ijadhadh.2010.07.005.
[7]
Abou-Hamda MM, Megahed MM, Hammouda MMI. Fatigue crack growth in double cantilever beam specimen with an adhesive layer. Eng Fract Mech 1998;60:605–14. doi:10.1016/S0013-7944(98)00018-6.
[8]
Banea MD, da Silva LFM. Static and fatigue behaviour of room temperature vulcanising silicone adhesives for high temperature aerospace applications. Statisches
Verhalten
und
Dauerfestigkeitsanalyse
von
vulkanisierten
Silikonklebstoffen für Luftfahrtanwendungen bei hohen Temperat. Materwiss 26
Werksttech 2010;41:325–35. doi:10.1002/mawe.201000605. [9]
Abdel Wahab MM. Fatigue in Adhesively Bonded Joints: A Review. ISRN Mater Sci 2012;2012:1–25. doi:10.5402/2012/746308.
[10]
Costa M, Viana G, da Silva LFM, Campilho RDSG. Environmental effect on the fatigue degradation of adhesive joints: A review. J Adhes 2017;93:127–46. doi:10.1080/00218464.2016.1179117.
[11]
Costa M, Viana G, da Silva L, Campilho R. Effect of humidity on the fatigue behaviour of adhesively bonded aluminium joints. Lat Am J Solids Struct 2017;14:174–87. doi:10.1177/1464420716645263.
[12]
Barros S De, Kenedi PP, Ferreira SM, Budhe S, Bernardino AJ. Influence of mechanical surface treatment on fatigue life of bonded joints. J Adhes 2017;93:599–612. doi:10.1080/00218464.2015.1122531.
[13]
Loureiro a. L, da Silva LFM, Sato C, Figueiredo M a. V. Comparison of the Mechanical Behaviour Between Stiff and Flexible Adhesive Joints for the Automotive
Industry.
J
Adhes
2010;86:765–87.
doi:10.1080/00218464.2010.482440. [14]
Boutar Y, Naïmi S, Mezlini S, da Silva LFM, Ben Sik Ali M. Characterization of aluminium one-component polyurethane adhesive joints as a function of bond thickness for the automotive industry: Fracture analysis and behavior. Eng Fract Mech 2017;177:45–60. doi:10.1016/j.engfracmech.2017.03.044.
[15]
Boutar Y, Naïmi S, Mezlini S, da Silva LFM, Hamdaoui M, Ben Sik Ali M. Effect of adhesive thickness and surface roughness on the shear strength of aluminium one-component
polyurethane
applications.
Int
J
adhesive Adhes
single-lap Sci
joints
Technol
for
automotive
2016;4243:1–17.
doi:10.1080/01694243.2016.1170588. [16]
Azari S, Papini M, Spelt JK. Effect of adhesive thickness on fatigue and fracture of toughened epoxy joints - Part I: Experiments. Eng Fract Mech 2011;78:153–62. doi:10.1016/j.engfracmech.2010.06.025.
[17]
Azari S, Papini M, Spelt JK. Effect of Surface Roughness on the Performance of Adhesive Joints Under Static and Cyclic Loading. J Adhes 2010;86:742–64. doi:10.1080/00218464.2010.482430. 27
[18]
Boutar Y, Naïmi S, Mezlini S, Ali MBS. Effect of surface treatment on the shear strength of aluminium adhesive single-lap joints for automotive applications. Int J Adhes Adhes 2016;67:38–43. doi:10.1016/j.ijadhadh.2015.12.023.
[19]
Kwon JW, Lee DG. The effects of surface roughness and bond thickness on the fatigue life of adhesively bonded tubular single lap joints. J Adhes Sci Technol 2000;14:1085–102. doi:10.1163/156856100743095.
[20]
Blanchard C, Chateauminois A, Vincent L. A new testing methodology for the assessment of fatigue properties of structural adhesives. Int J Adhes Adhes 1996;16:289–99. doi:10.1016/S0143-7496(96)00018-8.
[21]
Pereira AM, Ferreira JM, Antunes F V., Bártolo PJ. Study on the fatigue strength of AA 6082-T6 adhesive lap joints. Int J Adhes Adhes 2009;29:633–8. doi:10.1016/j.ijadhadh.2009.02.009.
[22]
Underhill PR, Rider AN, Duquesnay DL. The effect of warm water surface treatments on the fatigue life in shear of aluminum joints. Int J Adhes Adhes 2006;26:199–205. doi:10.1016/j.ijadhadh.2004.10.005.
[23]
Pascoe JA, Alderliesten RC, Benedictus R. Methods for the prediction of fatigue delamination growth in composites and adhesive bonds – A critical review. Eng Fract Mech 2013;112–113:72–96. doi:10.1016/j.engfracmech.2013.10.003.
[24]
Shenoy V, Ashcroft IA, Critchlow GW, Crocombe AD, Abdel Wahab MM. Strength wearout of adhesively bonded joints under constant amplitude fatigue. Int J Fatigue 2009;31:820–30. doi:10.1016/j.ijfatigue.2008.11.007.
[25]
Gupta VB, Drzal LT, Adams WW, Omlor R. An electron microscopic study of the morphology
of
cured
epoxy
resin.
J
Mater
Sci
1985;20:3439–52.
doi:10.1007/BF01113751. [26]
Shenoy V, Ashcroft IA, Critchlow GW, Crocombe AD, Wahab MMA. An investigation into the crack initiation and propagation behaviour of bonded singlelap joints using backface strain. Int J Adhes Adhes 2009;29:361–71. doi:10.1016/j.ijadhadh.2008.07.008.
28