The effect of temperature on the mechanical ...

51

The effect of temperature on the mechanical properties of adhesives for the automotive industry M D Banea1 and L F M da Silva2∗ 1 Instituto de Engenharia Mecânica (IDMEC), Porto, Portugal 2 Departamento de Engenharia Mecânica, Faculdade de Engenharia, Universidade do Porto, Porto, Portugal The manuscript was received on 28 May 2009 and was accepted after revision for publication on 3 December 2009. DOI: 10.1243/14644207JMDA283

Abstract: The application of adhesively bonded joints in structural components made of composite materials for automotive industry applications has increased significantly in recent years and provides many benefits that will ultimately lead to lighter-weight vehicles, fuel savings, and reduced emissions. The principal benefits are design flexibility, opportunity for part consolidation, and joining of dissimilar materials. While much work has been conducted in adhesive bonding for the aerospace industry, the automotive industry does not currently have a full portfolio of processes and methods for evaluating candidate adhesives for use in bonding structural automotive components. Aerospace techniques and materials are not generally applicable, since the automotive industry must be more cognizant of cost and high volume production. In this article, the performances of two different adhesive types, an epoxy and a polyurethane, have been studied through bulk specimen and adhesive joint tests. Results showed that the failure loads of both the bulk test and joint test specimens vary with temperature and this needs to be considered in any design procedure. Also, for the polyurethane adhesive, the single lap joint is sufficient to determine the adhesive shear strength. Keywords: adhesively bonded joints, automotive industry, temperature tests 1

INTRODUCTION

For automotive industry applications, the requirements for lightweight and durable parts for better fuel efficiency have led to an increasing use of fibrereinforced plastic, which have been used to replace metal automotive parts such as roofs, doors, and outer body panels. The traditional fasteners usually result in the cutting of fibres and hence the introduction of stress concentrations, both of which reduce structural integrity. The application of adhesively bonded joints in structural components made of reinforced composites has increased significantly in recent years and provides many benefits that will ultimately lead to lighter-weight vehicles, fuel savings, and reduced emissions. The principal benefits are design flexibility, opportunity for part consolidation, and joining of

∗ Corresponding

author: Departamento de Engenharia Mecânica,

(DEMec), Faculdade de Engenharia, Universidade do Porto (FEUP), Rua Dr. Roberto Frias, Porto 4200-465, Portugal. email: [email protected] JMDA283

dissimilar materials [1]. Additionally, adhesive bonding can result in stiffer assemblies and better load distribution regardless of the substrates. On the other hand, adhesively bonded joints are nowadays being used to substitute high-demanding welding technologies for materials such as aluminium, like the lightweight chassis of Lotus Evora, Morgan Supersports, and Aston Martin DB9. However, the composite materials have been restrictively applied to the automotive industry due to their low specific stiffness. To overcome this disadvantage, sandwich construction is frequently adopted instead of increasing material thickness. This type of construction consists of two face sheets that carry the bending stresses and a low-density core that resists the shear stress. High energy absorption during impact makes these sandwich structures attractive to designers. Several typical joint configurations for sandwich panels (foam core) are illustrated in Fig. 1. The influence of temperature on the strength of adhesive joints is an important factor to consider in the their design. The most significant factors that determine the strength of an adhesive joint when used over a wide temperature range are the coefficients of Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

52

M D Banea and L F M da Silva

Fig. 1 Typical joint configurations for sandwich panels

thermal expansion (CTE) (especially when compared to the CTE of the substrates) [2], and different adhesive mechanical properties with temperature [3–5]. Studies that present experimental results of adhesive joints with structural adhesives (especially epoxies) as a function of temperature generally show a decrease in strength with increasing and decreasing temperatures [3, 6–8]. At high temperatures the cause is the low adhesive strength, while at low temperatures it is the high thermal stresses. Adams et al. [6] studied the performance of single lap joints (SLJs) with epoxy adhesives at low and room temperatures (RTs). They investigated the effects of adherend mismatch, shrinkage, and adhesive properties on the stress state of lap joints and found that the stresses caused by adhesive shrinkage have much less effect on the lap joint strength than those generated by the adherend thermal mismatch. Owens and Lee-Sullivan [9] tested SLJs with a rigid and a flexible epoxy adhesive at RT and at −40 ◦ C in quasi-static conditions. They studied stiffness loss due to crack growth in composite-toaluminium joints. Results showed that the joint stiffness is more affected by the response of the adherends to the test temperature than by the modulus of the thin adhesive layer. The properties of adhesives and sealants over the range of service temperatures need to be studied for each type of application. For example, the adhesive joints used in automotive industry need to withstand temperatures between −40 and 80 ◦ C [10], so the adhesives’ stress–strain data must be characterized over the range of these temperatures. The adhesives investigated here find their main application in the automotive and bus industry. For example, the flexible adhesive Sikaflex-552 is used in elastic joints and seals in vehicle construction, while the adhesive Araldite AV118 is used in more structural bonds. Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

In this study, bulk specimens of cured adhesives were produced according to the French standard NF T 76-142 [11] and tested in tension at RT, −40 ◦ C, and 80 ◦ C, in order to obtain a strength profile of the adhesives over this temperature range. The standard thick adherend shear test (TAST) was performed in order to measure the shear properties of the adhesives according to standard ISO 11003-2:1993 [12]. The TAST is preferred for determining design parameters as the thick, rigid adherends reduce (but not eliminate) the peel stresses [13]. In addition, SLJs were fabricated and tested to assess the adhesives’ performance in a joint. The influence of temperature on the lap shear strength of the adhesives was investigated.

2 2.1

EXPERIMENTAL DETAILS Adhesives selected

Two adhesives were chosen for this study: Araldite AV118, a one-component epoxy adhesive supplied by Huntsman and Sikaflex-552, and a one-component polyurethane hybrid adhesive supplied by Sika Portugal (Sikaflex® hybrid technology combines the high performance of Sikaflex polyurethane systems with silanes, which gives great adhesion with little or no surface preparation, removing the necessity to use a primer). Structural adhesives, such as Araldite AV118, have high glass transition temperatures and high elastic modulus, but low extensions to failure. Flexible adhesives, such as Sikaflex-552, have low glass transition temperatures and low elastic modulus, but high extensions to failure. However, the advantageous properties of flexible adhesives in sustaining large strains and distributing peel forces more evenly on the bonded substrates lead to their use for structural JMDA283

The effect of temperature on the mechanical properties of adhesives

joining applications in automotive industry. On the other hand, combinations of rigid and flexible adhesive are also possible. This would be particularly beneficial for bonded adherends with dissimilar CTE or with different stiffnesses [4]. 2.2 2.2.1

Specimens manufacture Bulk tensile specimens

Thin sheets of Araldite AV118 adhesive were produced by curing the AV118 adhesive between steel plates of a mould with a silicone rubber frame (Fig. 2(a)) according to the French standard NF T 76-142, which were hot pressed (2 MPa) for 60 min at 120 ◦ C (according to the manufacturer’s recommended cure schedule). The silicone rubber frame stops the adhesive from flowing out and the pressure applied creates a good surface finish. The dimensions of the adhesive plate after cure were 150 mm × 45 mm, with a thickness of 2 mm (see Fig. 3(a)), which corresponds to the internal dimensions of the silicone rubber frame. Dogbone specimens 2 mm thick and 10 mm wide were machined from the bulk sheets plates (Fig. 3(a)). The geometry of the dogbone tensile specimens used (BS 2782 standard) is shown in Fig. 4(a). AV118 dogbone tensile specimens, before and after tests, can be seen in Fig. 3(b). The fully cured flexible Sikaflex-552 adhesive is very soft and machining to the right dimensions was not

Fig. 2

feasible. Therefore, it was decided that it would be better to use a moulding technique to avoid any problems associated with the cutting of the specimens and to produce the best possible specimens for tensile testing. The metallic mould designed for this purpose is shown in Fig. 2(b). It consists of three individual parts: (a) the base part, which is used as the support; (b) the middle part that determines the shape and the thickness of the specimen; (c) the top part that closes the mould. To ensure that the adhesive does not bond to the mould, three layers of release agent were applied to all the parts, prior to adhesive application. For the application of the release agent, the metallic moulds were heated to 60 ◦ C. The three parts were fitted together with screws. The Sikaflex-552 dogbone specimens were cured at RT, following the manufacturer’s suggested curing conditions (25 ◦ C and 50 per cent relative humidity (RH)) (Fig. 3(c)). For a fast and complete curing process the specimens were removed from the mould after 1 week and left at RT for a period of three more weeks. In this way, the diffusion of water from all sides into the bulk of the samples was allowed to promote complete curing of the specimen. The geometry of the dogbone tensile specimens (2 mm thick and 10 mm wide) used correspond to BS 2782 standard and is shown in Fig. 4(b).

Moulds for producing bulk specimens

Fig. 3 JMDA283

53

Bulk specimens Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

54


Fig. 4 Tensile test specimen geometry bulk specimens (dimensions in mm)

2.2.2

Thick adherend shear test

For the TAST, steel substrates of dimensions 110 × 25 × 12 mm3 (Fig. 5) were used. The joint surfaces were grit blasted and degreased with acetone prior to the application of the adhesive. The bondline thickness was nominally 0.7 mm and the length of the overlap test section was 5 mm. Two spacers (1.5 mm thick) were inserted in the gaps between the adherends after the application of the adhesive and prior to curing in order to provide the necessary spacing between the

Fig. 5

two adherends. These spacers were removed after the adhesive was cured. A mould with spacers for correct alignment of the specimens was used and is shown in Fig. 6. Adhesive AV118 was cured in a hot press, following the manufacturer’s suggested curing conditions (60 min at 120 ◦ C), and adhesive Sikaflex-552 was cured at RT (25 ◦ C and 50 per cent RH) for 1 week. Prior to testing, each specimen must be dimensioned for use in calculations and to assure conformity to the dimension standards set out in the ISO

Standard TAST specimen (dimensions in mm)

Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

JMDA283


55

and then the joints were left under pressure (0.5 MPa) for 24 h at RT in a hydraulic press. They were then removed from the mould and left for another 10 days to fully cure the adhesive, following the manufacturer’s suggested curing conditions (25 ◦ C and 50 per cent RH). After the end of the curing process, any excess adhesive was carefully removed. The glue line thickness was 0.2 mm and the length of the overlap was 25 mm. In order to achieve that, spacers with thicknesses of 2.2 mm, in order to provide the necessary spacing between the two adherend halves, were used. Fig. 6

Mould for TAST specimens fabrication

2.3 Test method 11003-2:1993 guideline. Measurements for each specimen were taken and recorded for the width, length, and thickness of the bondline. 2.2.3

Single lap joints

For Araldite AV118 adhesive, high-strength steel substrates of dimensions 107.5 × 25 × 2 mm3 were used in order to avoid plastic deformation of the adherends. The joint surfaces were grit blasted and degreased with acetone prior to the application of the adhesive. The substrates were bonded and then the specimens were cured in a hot press following the manufacturers’ suggested curing conditions (60 min at 120 ◦ C). The cure conditions for Araldite AV118 adhesive were the same for all specimens, in accordance with the manufacturer’s recommended cure schedule, in order to assure similar mechanical properties (for comparing data from different mechanical tests it is essential that differences in properties due to the cure state are minimized). A mould with spacers for correct alignment of the substrates was used. The bondline thickness was controlled using packing shims. The bondline thickness was 0.2 mm and the length of the overlap was 12.5 mm. The geometry of the lap shear joint specimens used is shown in Fig. 7. In the case of Sikaflex-552 adhesive, mild steel substrates (the adherends were considered to be almost infinitely rigid in comparison with the low modulus polyurethane adhesive) of dimensions 110 × 25 × 2 mm3 were used. The joint surfaces were grit blasted and degreased with Sika cleaner prior to the application of the adhesive. The substrates were bonded

Fig. 7 JMDA283

2.3.1

Tensile tests

AV118 specimens were tested in tension using a universal testing machine Instron model 4208, which has a maximum load capacity of 300 kN, under a constant crosshead rate of 1 mm/min. A load cell of 10 kN was used. An Instron extensometer (50 mm gauge length) was used to record the displacement. A pair of clamps was used to grip the specimens (Fig. 8(a)). For the hightemperature tests, an environmental chamber of the machine was used to reach the desired temperature (80 ◦ C). For the −40 ◦ C temperature tests, the chamber of the machine was cooled with CO2 . At least three ‘dogbone’ tensile specimens were tested to failure at each temperature. The Sikaflex-552 adhesive specimens were tested only at RT in tension using a testing machine TIRA model 2705, under a constant crosshead rate of 5 mm/min. A load cell of 5 kN was used. Three ‘dogbone’ tensile specimens were tested to failure. The TIRA 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 extensometry, 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 fracture occurs. However, the non-contacting strain measurement techniques offer the great advantage of measuring the actual strain on the gauge length of

Lap joint specimen geometry (dimensions in mm) Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

56


Fig. 8

(a) An AV118 dogbone specimen clamped in the Instron machine with an extensometer attached and (b) a Sikaflex-552 dogbone specimen clamped in the Tira machine

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. 8(b)), 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. The digital images were then analyzed using an image processing and analysis software and the strain was extracted for each specimen. 2.3.2

Fig. 9

MTS extensometer

Thick adherend shear test

TAST tests were performed at RT on an MTS servohydraulic machine, model 312.31, at a constant crosshead rate of 0.1 mm/min. For load measurements, 10 per cent of the capacity of the load cell (25 kN) was used. The specimens were fixed in the machine through a pair of clamps. The correct alignment of the specimens was assured by a pin. The displacement was measured with two methods: a 25-mm-long MTS extensometer (Fig. 9) and a non-contact method (video microscopy). As the extensometer is mounted in the metallic substrate, the extensometer measures not only the displacement of the adhesive, but also the displacement of the adherend. Therefore, it is necessary to apply a correction to the measured displacements. At the same time video microscopy (Fig. 10) was used to record the displacements, which gives only the adhesive displacement. The strains were calculated using the spatial correlation method developed by Chousal and Gomes [14]. da Silva et al. [15] showed that the steel deformation can be neglected in the case of flexible adhesives, so Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

Fig. 10 Video microscopy

that the Sikaflex adhesive displacement was measured only by the MTS extensometer method. At least five TAST specimens were tested to failure for each adhesive. 2.3.3

SLJ tests

Testing was conducted at RT at a constant displacement rate of 1 mm/min using an MTS 312.31 servohydraulic machine. Loads and displacements to failure were recorded. SLJs were tested at −40 and 80 ◦ C using a universal testing machine Instron model 4208, under a constant crosshead rate of 1 mm/min. A load cell JMDA283

3

57

4.0

80

3.5

70

3.0

60

2.5

50

2.0

40

1.5

30

1.0

20

0.5

10

0.0

Tensile strength (MPa)

of 10 kN was used for AV118 joints and a 5 kN load cell for Sikaflex-552 adhesive joints. An Instron extensometer (of 50 mm gauge length) was used to record the adhesive displacement. For the high- and low-temperature tests, the environmental chamber of the machine was used to reach the desired temperature (80 and −40 ◦ C). Three joints were tested to failure at each temperature. For each joint tested, load–displacement curves were produced.

Young's modulus (GPa)


0 -40

23 60 Temperature (ºC) Young's modulus

80

Strength

Fig. 11

Adhesive AV118 average Young’s modulus and tensile strengths as a function of temperature

Fig. 12

Representative AV118 adhesive tensile stress– strain curves as a function of temperature

RESULTS AND DISCUSSION

3.1 Tensile tests AV118 dogbone specimens were tested at RT (23 ◦ C), −40 ◦ C, and 80 ◦ C. For each specimen, a tensile stress– strain curve was produced based on load, displacement values, and specimen dimensions. From the stress–strain curve, the elastic modulus, ultimate tensile strength, and maximum tensile strain were calculated and the results are presented in Table 1. The values for Young’s modulus were calculated from the tangent to the tensile stress–strain curve at the origin (a polynomial approximation of the curve was made). The Instron extensometer could not record the displacements and strain until the fracture of the specimens at 80 ◦ C because of the extensometer length limitation (the adhesive became more ductile resulting in higher tensile displacements), which did not allow the measurements of the strain until failure. Subsequently, one specimen was tested at 60 ◦ C. Young’s modulus, ultimate tensile strength, and maximum tensile strain were determined as 2455 MPa, 44.8 MPa, and 10.5 per cent, respectively. The variation of Young’s modulus and tensile strength as a function of temperature are presented in Fig. 11. Representative adhesive tensile stress–strain curves as a function of temperature can be seen in Fig. 12. The data obtained show a decrease in AV118 adhesive strength with increasing temperature and an increase in the ductile response of the adhesive. As the temperature increases, the adhesive becomes more ductile, resulting in more tensile displacement (strain) Table 1 Tensile modulus, strength, and strain data for AV118 specimens tested at RT (23 ◦ C), −40 ◦ C, and 80 ◦ C

to failure and less strength (Figs 11 and 12). Also, as the temperature decreases, the adhesive becomes more brittle and has more apparent tensile strength, but less strain. The Sikaflex-552 dogbone specimens were tested at RT. For each specimen, a tensile stress–strain curve was produced based on load, displacement values, and specimen dimensions. From the stress– strain curve, the elastic modulus and ultimate tensile strength were calculated and the results are presented in Table 2. The values for Young’s modulus were calculated from the tangent to the tensile stress– strain curve at the origin (a polynomial approximation of the curve was made). Representative Sikaflex-552 adhesive tensile stress–strain curve can be seen in Fig. 13. 3.2 Thick adherend shear test In Fig. 14(a), a characteristic shear stress–strain curve of AV118 adhesive measured with the MTS extensometer is shown. At the same time, a video microscopy was used to record the displacements, which gives only the

Temperature (◦ C)

Young’s modulus (MPa)


Tensile strain (%)

Table 2 Tensile modulus, strength, and strain data for Sikaflex-552 specimens tested at RT

RT −40 60 80

3010 ± 206 3617 ± 150 2455 1852 ± 99

65.9 ± 5.7 69.7 ± 14.5 44.8 29.7 ± 1.4

5.90 ± 0.24 2.46 ± 0.87 10.5 –

Temperature (◦ C)

Young’s modulus (MPa)


Tensile strain (%)

RT

4.17 ± 0.5

2.5 ± 0.25

231 ± 28

JMDA283


58


Table 3

3.0

Stress (MPa)

2.5 2.0 1.5 1.0 0.5

Shear modulus and strength data for AV118 tested at RT

Shear modulus, G (steel + adhesive curve) (MPa)

Shear modulus, G (adhesive curve) (MPa)

Shear strength (MPa)

Shear strain (%)

811 ± 119

1095 ± 19

37 ± 1.4

27.9 ± 4.3

0.0 0.0

0.5

1.0

1.5

2.0

2.5

Strain

Fig. 13 Typical Sikaflex-552 tensile stress–strain curve

adhesive displacement. A characteristic shear stress– strain curve of AV118 adhesive measured by the two methods (MTS extensometer and video microscopy) is shown in Fig. 14(b). Shear modulus, shear strength, and maximum shear strain of AV118 tested at RT (23 ◦ C) can be found in Table 3. The shear modulus is related to Young’s modulus by the following equation G=

E 2(1 + ν)

(1)

where G is the shear modulus and ν is the Poisson’s ratio. From equation (1) the value of Poisson’s ratio can be calculated v=

E −1 2G

(2)

3.3

SLJ tests

A summary of maximum load and average lap shear strength for AV118 SLJs tested at RT, −40 ◦ C, and 80 ◦ C are presented in Table 5. The average lap shear strength is given by σ =

40

40

35

35

30 25 20 15 10

25 20 15 10 5

0

0

0,05

0,1

0,15

0,2

0,25

0,3

0,35

Shear strain

a)

(3)

30

5 0

P bL

where P is the maximum load, b is the joint width, and L is the joint overlap length. Average lap shear strength of AV118 SLJ as a function of temperature is presented in Fig. 17. The lap shear strength of the adhesive joints tested at 80 ◦ C was approximately 30 per cent less than that of the specimens tested at RT, while data obtained from tests at −40 ◦ C showed a decrease of the lap shear strength of the adhesive by approximately 10 per cent. Generally, as the temperature is raised, the adhesive strength decreases and the ductility increases. There is

Shear Stress (MPa)

Shear Stress (MPa)

Applying equation (2) and using G and E experimental, the value of ν is 0.37. As the value of Poisson’s ratio typically varies between 0.3 and 0.5, this means that the shear modulus measured with the TAST is reasonable. Typical shear stress–strain curves for the Sikaflex552 adhesive tested at RT are shown in Fig. 15. From the shear stress–strain curve, the shear modulus and shear strength were calculated. In general, elastomeric materials exhibit non-linear stress–strain behaviour and the definition of the modulus is very difficult. The values for shear modulus were calculated from the tangent to the shear stress–strain curve at the origin (a polynomial approximation of the curve was made).

The shear modulus, shear strength, and strain data for Sikaflex-552 are presented in Table 4. For incompressible materials, usually ν is assumed to be equal to 0.5. Applying equation (2) and using G and E experimental for Sikaflex-552, the value of ν is 0.60. The shear modulus predicted from bulk tension tests (applying equation (1) and ν = 0.5) is slightly higher (1.39) than the experimentally measured from TAST tests (1.30). This is probably due to the different manufacturing technique of the specimens used for these tests. Typical failure modes of adhesives in TAST specimens are presented in Fig. 16. The failure is a cohesive/adhesive mixed-mode failure.

0,4

Adhesive

0

0,05

0,1

0,15

Steel+adhesive

0,2

0,25

0,3

0,35

Shear Strain

b)

Fig. 14 Typical AV118 shear stress–strain curve measured with (a) MTS extensometer and (b) by the two methods: MTS extensometer and video microscopy Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

JMDA283


59

Shear Stress (MPa)

3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.0

0.5

1.0

1.5 sample 1

2.0 Strain

2.5

sample 2

3.0

3.5

4.0

sample 3

Fig. 15 Typical shear stress–strain curves of Sikaflex-552 TAST specimens tested at RT Shear modulus and Sikaflex-552 tested at RT

strength

data

for

Shear modulus G (MPa) Shear strength (MPa) Shear strain (%) 1.30 ± 0.12

2.39 ± 0.18

330 ± 16

a temperature when the adhesive starts to behave like a rubber: this is called the glass transition temperature Tg . In the region of this temperature, there is a drop in the elastic modulus and in the strength. Below Tg it is difficult to say at which temperature the lap shear strength is the highest as there are two factors to consider: the ductility and the strength. Figure 18 shows schematically the lap shear strength behaviour as a function of the adhesive bulk properties. The lap shear strength increases with the adhesive ductility up to the best compromise between the ductility and the bulk strength. According to Fig. 17, the temperature corresponding to the best combination strength–ductility for AV118 adhesive is the RT. AV118 epoxy adhesive has a Tg of approximately 109 ◦ C (determined with the dynamic mechanical thermal analysis method) and is in the glassy state at RT. Decline of bond strength with rise of temperature may also be explained by adhesive failure (see section 3.4). A summary of maximum load and average lap shear strength for Sikaflex-552 SLJs tested at RT, −40 ◦ C, and 80 ◦ C are presented in Table 6. Average lap shear strength of Sikaflex-552 SLJs as a function of temperature is presented in Fig. 19.

Fig. 16 JMDA283

With an increase in temperature, a slight decrease in the lap shear strength occurs because of the decrease in adhesive strength. The lap shear strength of the adhesive joints tested at 80 ◦ C is approximately 20 per cent less than that of the specimens tested at RT. Data obtained from tests at −40 ◦ C showed an increase of the lap shear strength of the adhesive by approximately 115 per cent, approximately two times higher than SLJs tested at RT. This is explained by the fact that

Table 5

Maximum load and average lap shear strength for AV118 SLJs tested at RT, −40 ◦ C, and 80 ◦ C

Temperature (◦ C)

Maximum load (kN)

Average lap shear strength (MPa)

RT −40 80

11.8 ± 0.1 10.6 ± 0.2 8.0 ± 0.5

37.7 ± 0.2 33.9 ± 0.6 25.7 ± 1.8


Table 4

40 35 30 25 20 15 10 5 0 -60

Fig. 17

-40

-20

0

20 40 Temperature (ºC)

60

80

100

Average lap shear strength of AV118 as a function of temperature

Failure mode in TAST specimens: (a) AV118 and (b) Sikaflex-552 Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

60


Fig. 18

Table 6

Lap shear strength behaviour as a function of ductility and strength Maximum load and average lap shear strength for Sikaflex-552 SLJs tested at RT, −40 ◦ C, and 80 ◦ C Maximum load (kN)


RT −40 80

2.02 ± 0.19 4.36 ± 0.04 1.58 ± 0.09

3.23 ± 0.31 6.98 ± 0.06 2.54 ± 0.15

8 7

3.3.1

Failure load prediction

The failure load of SLJs with ductile adhesives (case of Sikaflex-552) can be predicted using the simple design methodology proposed by Adams et al. [16], based on the shear stress of the adhesive. The load corresponding to the total plastic deformation of the adhesive (global yielding) is given as P = τy × b × L

6

(4)

5 4 3 2 1 0 -60

Fig. 19

-40

-20

0

20 40 Temperature (ºC)

60

80

100

Average lap shear strength of Sikaflex-552 SLJs as a function of temperature

polyurethane adhesives have low glass transition temperatures (Tg = −60 ◦ C for Sikaflex-552, data provided by supplier). They remain ductile and their strength increases at low temperatures that lead to a higher joint strength. For Sikaflex-552 adhesive is difficult to say at which temperature the lap shear strength is the highest as tests around its Tg (−60 ◦ C) have not been performed. However, for the range of temperature tested here (−40 to 80 ◦ C), the temperature corresponding to the optimum combination of strength/ductility is −40 ◦ C. In summary, the glass transition temperature Tg of the adhesive is a key parameter in the testing of adhesive joints. While the AV118 epoxy adhesive is in the Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

where P is the failure load of the adhesive, τy is the shear yield strength of the adhesive, b is the joint width, and L is the overlap length. In Fig. 20, experimental and predicted failure loads (with values of shear stress obtained from TAST) of the SLJs are shown. It can be seen that the simple criterion adopted for the joints gives failure loads that compare quite well with the experimental results. Sikaflex 552 Eq. 4 Sikaflex 552 Exp. AV118 Eq. 5 AV 118 Exp.

18 16 Failure load (kN)

Average lap shear Strength (MPa)

Temperature (◦ C)

glassy state at RT (Tg of approximately 109 ◦ C), the polyurethane Sikaflex-552 adhesive is in a rubbery state at RT (Tg of −60 ◦ C), so that they have different behaviour in adhesively bonded joints when tested as a function of temperature. Also, for the joints with Sikaflex-552 flexible adhesive tested in this work, the average lap shear stress in the joints at failure in quasi-static loading is very similar to the measured shear strength with the TAST (For Sikaflex-552 from TAST, τr = 2.39 MPa and from SLJs τav = 2.27 MPa). This indicates that the shear stresses in the joints are essentially uniformly distributed whether a TAST or SLJ specimens are used. In other words, the SLJ can be used to determine the shear strength of flexible adhesives, contrarily to stiff and rigid adhesives like epoxies. The joint strength depends not only on the adhesive shear strength but also on its ductility. While a ductile adhesive enables to redistribute the load and make use of the less stressed parts of the overlap, a stiff adhesive generate high localized stresses at the joint ends with very little stress carried in the central region of the overlap.

14 12 10 8 6 4 2 0 0

10

20

30

40

50

60

Overlap length (mm)

Fig. 20

Experimental and predicted failure loads of SLJs JMDA283


Fig. 21

Fig. 22

Failure mode of AV118 SLJs tested at (a) RT, (b) −40 ◦ C, and (c) 80 ◦ C

Failure mode of Sikaflex-552 SLJs tested at (a) −40 ◦ C, (b) RT, and (c) 80 ◦ C

For joints with ductile adhesives, the failure load is given by the load that causes adhesive global yielding along the overlap. This criterion works reasonably well provided the failure shear strain of the adhesive is more than 20 per cent, which is the case of Sikaflex-552 adhesive used in the present study. However, for brittle adhesives (AV118), this methodology is not applicable [15]. For joints with a brittle adhesive, the Volkersen’s model [17] is used and the failure occurs when the maximum shear stress at the ends of the overlap exceeds the shear strength of the adhesive. The following equation was used P = τr

2bl sinh(λl) λl[1 + cosh(λl)]

(5)

where λ2 =

G ta

2 Ets

ta is the adhesive thickness, ts is the substrate thickness, G is the adhesive shear modulus, and E is the adherend Young’s modulus. The Volkersen’s criterion works reasonably well for the brittle adhesive AV118. 3.4

Failure modes

After the tests, the failure modes of the specimens were evaluated visually. In Fig. 21, typical failure modes for AV118 SLJs are presented. Failure of the adhesive occurs in regions of JMDA283

61

maximum stress or strain concentration and results in cracks that run through the adhesive (at right angles to the direction of the maximum tensile stress or strain). The failure is partially cohesive failure close to the adherend–adhesive interface (mixed-mode failure) for SLJs tested at RT (Fig. 21(a)). For AV118 SLJs tested at −40 ◦ C, the failure was cohesive failure mode (Fig. 21(b)) and for AV118 SLJs tested at 80 ◦ C, the failure mode was adhesive failure (Fig. 21(c)). For Sikaflex-552 SLJs, the failure was cohesive within the adhesive in all cases. However, the appearance of the failure bond surfaces alters with temperature as can be seen in Fig. 22. The failure surfaces at −40 ◦ C (Fig. 22(a)) show little adhesive deformation, indicating that the adhesive becomes less ductile, while the failure surface of the adhesives tested at 80 ◦ C (Fig. 22(c)) shows an increase in adhesive deformation which is a sign of more ductility.

4

CONCLUSIONS

In this article, two different adhesive types, an epoxy and a polyurethane, were studied through bulk specimen and adhesive joint tests. The following conclusions can be drawn. 1. Tensile data obtained from tests at RT showed that the AV118 adhesive has a high strength (66 MPa) and stiffness (E = 3.01 GPa). Large increases in ductility (strain to failure) and reductions in strength were measured at elevated temperature (80 ◦ C). Tensile data obtained from tests at −40 ◦ C showed Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications

62

2.

3.

4.

5.

6.


that, as the temperature decreases, the AV118 adhesive becomes more brittle and has more apparent tensile strength, but less strain. Tensile data obtained from tests at RT showed that the Sikaflex-552 adhesive has a low strength and stiffness but sustains large displacements. The standard test TAST was performed for each adhesive in order to measure the shear properties of the adhesive. SLJs of AV118 and Sikaflex-552 adhesive were tested at RT, −40 ◦ C, and 80 ◦ C. Tests results showed that the lap shear strength of the adhesive is affected by variation of temperature. For AV118, the temperature corresponding to the highest strength is RT whenever for Sikaflex-552 it is −40 ◦ C. The failure loads of both the bulk test and joint test specimens vary with temperature and this needs to be considered in any design procedure. The AV118 adhesive offers high strength and stiffness. The Sikaflex-552 adhesive provides flexible and resilient joints. It also sustains large strains and distributes peel forces more evenly on the bonded substrates. These properties are leading to their use for structural joining applications in the automotive industry.

4

5

6

7

8

9

10 11

ACKNOWLEDGEMENTS The authors thank the Portuguese Foundation for Science and Technology for supporting the work presented here, through the research project PTDC/EMEPME/67022/2006 and the European Commission, VI Framework Programme (project LITEBUS TST5-CT2006-031321), and Sika Portugal for supplying the Sikaflex adhesive.

12

13

14

© Authors 2010 REFERENCES 1 Banea, M. D. and da Silva, L. F. M. Adhesively bonded joints in composite materials: an overview. Proc. IMechE, Part L: J. Materials: Design and Applications, 2009, 223(L1), 1–18. DOI: 10.1243/14644207JMDA219. 2 da Silva, L. F. M. and Adams, R. D. Stress-free temperature in a mixed-adhesive joint. J. Adhesion Sci. Technol., 2006, 20(15), 1705–1726. 3 Adams, R. D. and Mallick, V. The effect of temperature on the strength of adhesively-bonded


15

16

17

composite-aluminum joints. J. Adhesion, 1993, 43(1–2), 17–33. da Silva, L. F. M. and Adams, R. D. Adhesive joints at high and low temperatures using similar and dissimilar adherends and dual adhesives. Int. J. Adhesion Adhes., 2007, 27(3), 216–226. Deb, A., Malvade, I., Biswas, P., and Schroeder, J. An experimental and analytical study of the mechanical behaviour of adhesively bonded joints for variable extension rates and temperatures. Int. J. Adhesion Adhes., 2008, 28(1–2), 1–15. Adams, R. D., Coppendale, J., Mallick, V., and AlHamdan, H. The effect of temperature on the strength of adhesive joints. Int. J. Adhesion Adhes., 1992, 12(3), 185–190. Kang, S.-G., Kim, M.-G., and Kim, C.-G. Evaluation of cryogenic performance of adhesives using compositealuminum double-lap joints. Compos. Struct., 2007, 78(3), 440–446. Banea, M. D. and da Silva, L. F. M. Mechanical characterization of flexible adhesives. J. Adhesion, 2009, 85(4), 261–285. Owens, J. F. P. and Lee-Sullivan, P. Stiffness behaviour due to fracture in adhesively bonded composite-toaluminum joints, part 2: experimental. Int. J. Adhesion Adhes., 2000, 20(1), 47–58. Cognard, P. Collage des composites. Secteurs routier et ferroviaire, BM 7 627, 2003, pp. 1–7. da Silva, L. F. M., Adams, R. D., and Gibbs, M. Manufacture of adhesive joints and bulk specimens with hightemperature adhesives. Int. J. Adhesion Adhes., 2004, 24(1), 69–83. ISO 11003-2 1993 (E). Adhesives – determination of shear behaviour of structural bonds, part 2: thick-adherend tensile-test method, 1993. Vaughn, L. F. and Adams, R. D. Test methods for determining shear property data for adhesives suitable for design part 3: the thick-adherend shear test method, ADH1 report no. 8, NPL, 1996. Chousal, J. A. G. and Gomes, J. F. (Ed. S. A. Meguid) In Proceedings of the M2D, Third International Conference on Mechanics and Materials in Design, Orlando, FL, 2000, pp. 269–270 (University of Toronto). da Silva, L. F. M., Silva, R. A. M. D., Chousal, J. A. G., and Pinto, A. M. G. Alternative methods to measure the adhesive shear displacement in the thick adherend shear test. J. Adhesion Sci. Technol., 2008, 22(1), 15–29. Adams, R. D., Comyn, J., and Wake,W. C. Structural adhesive joints in engineering, 2nd edition, 1997 (Chapman and Hall, London). Volkersen, O. Die nietkraftoerteilung in zubeanspruchten nietverbindungen mit konstanten loschonquerschnitten, Luftfahrtforschung, 1938, 15, 41–47.

JMDA283