Mode II fracture toughness of a brittle and a ductile

adhesive as a function of the adhesive thickness

Lucas F M da Silva1,∗, F A C R G de Magalhães1, F J P Chaves2, M F S F de

Moura1
1
Departamento de Engenharia Mecânica, Faculdade de Engenharia da Universidade do Porto,

Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

Abstract

The main goal of this study was to evaluate the effect of the thickness and type of

adhesive on the mode II toughness of an adhesive joint. Two different adhesives were

used, Araldite AV138/HV998 which is brittle and Araldite 2015 which is ductile. The

end notched flexure test was used to determine the mode II toughness because it is

commonly known to be the easiest and widely used to characterize mode II fracture.

The specimen was placed over a fixture in contact with two extremities and in the

mean point a load is applied to promote flexure bending which will propagate the

crack through the bondline. The main conclusion is that the energy release rate for

AV138 does not vary with the adhesive thickness whereas for Araldite 2015, the

toughness in mode II increases with the adhesive thickness. This can be explained by

the adhesive plasticity at the end of the crack tip.

∗
Corresponding author. Tel: +351225081706. Fax: +351225081445. Email: lucas@fe.up.pt

1
Keywords

Epoxy; Brittle adhesive; Ductile adhesive; End notched flexure test; Mode II fracture

toughness; adhesive thickness.

1. Introduction

Adhesively bonded joints were initially designed using a continuum mechanics

approach. The maximum principal stress was proposed for very brittle materials

whose failure mode is normal to the direction of maximum principal stress [1-2].

However, because of the singularity of stresses at the re-entrant corners of joints, the

stresses depend on the mesh size used and how close to the singular points the stresses

are taken. Therefore, care must be taken when using this criterion. When ductile

adhesives are used, criteria based on maximum stress are not appropriate because

such joints can still carry large loads after adhesive yielding. For ductile adhesives,

Adams and Harris [2] used maximum principal strain as the failure criterion for

predicting joint strength. Hart-Smith [3] proposed that the maximum adhesive shear

strain might be used as a failure criterion when plastic deformation of the adhesive

occurred. da Silva et al. [4] implemented this criterion into a commercial package.

Other analyses go beyond that of Hart-Smith, by taking into consideration both shear

and peel contributions to plasticity, such as that of Adams and Mallick [5]. More

recently, da Silva et al. [6-7] have shown for single lap joints that the maximum shear

strain criterion is very accurate for ductile adhesives.

Continuum mechanics assumes that the structure and its material are continuous.

Defects or two materials with re-entrant corners obviously violate such an

assumption. Consequently, continuum mechanics gives no solution at these singular

2
points because of the stress or strain singularities. Cracks are the most common

defects in structures, for which the method of fracture mechanics has been developed.

In fracture mechanics (FM), it is well accepted that stresses calculated by using

continuum mechanics are singular (infinite) at the crack tip. Although FM is mainly

used for dealing with sharp cracks, angular wedged notches are also of practical

importance. The use of a generalized stress-intensity factor, analogous to the stress-

intensity factor in classical FM, to predict fracture initiation for bonded joints at the

interface corners has been investigated [8-10]. Damage mechanics has been used to

model the progressive damage and failure of a pre-defined crack path [11-14]. The

damage is confined to a zero volume line or a surface is often referred to as a cohesive

zone model (CZM). A CZM simulates the fracture process, extending the concept of

continuum mechanics by including a zone of discontinuity modelled by cohesive

zones, thus using both strength and energy parameters to characterize the debonding

process. This allows the approach to be of much more general utility than

conventional fracture mechanics.

In order to apply a fracture mechanics or damage mechanics approach, it is necessary

to have the fracture toughness of the material. The toughness varies with the type of

loading, i.e., mode I, II, III and mixed. Most of the data available in the literature is

for the fracture toughness in mode I using the double cantilever beam. However,

adhesive joints are also loaded in mode II and under mixed mode. For the

determination of the toughness in mode II there are various test methods available

(Figure 1): the end notched flexure (ENF) test, the end loaded split (ELS) test and the

four-point notched flexure (4ENF) test. The ELS test presents large displacements and

is sensitive to the camping device. The 4ENF is more sophisticated but has problems

of friction due to the loading mode in the pre-crack region. The easier and probably

3
most common testing method for mode II is the ENF test. In the ENF test, the

specimen is simply supported at the ends and a load is applied in the middle of the

specimen causing a shear mode loading in the adhesive.

It is known that the adhesive toughness varies with the adhesive thickness, especially

with ductile adhesives because of the constraining effects of the adherends. The

thickness of the adhesive layer contributes for the joint behaviour. Thus it should be

taken into account and thoroughly studied. Boscom et al. [15-16] found that fracture

energy is maximized when the adhesive layer thickness equals the diameter of the

fracture process zone (FPZ) ahead of the crack tip. Kinloch and Shaw [17] showed

that the FPZ played an important role in enhancing GIC of the adhesive joint.

Daghyani et al. [18-19] found a transition in the fracture process from a cohesive to an

interfacial adhesive failure for thin layers. Lee et al. [20] found that as bond thickness

decreases, the fracture energy either decreases monotonically, or increases, peaks, and

then decreases rapidly. Most of the results in the literature concerning the effect of the

adhesive thickness is for mode I, but little is available concerning mode II, which

should be the main loading mode in adhesive joints.

The main objective of the present study was to measure the mode II fracture

toughness of two types of adhesive (brittle and ductile) using the ENF test as a

function of the adhesive thickness.

4
2. Experimental details

2.1 Materials

Two adhesives were selected, a very stiff and brittle epoxy (AV138/HV998 from

Huntsman, Salt Lake City, UT) used in aerospace applications, and a more flexible

and ductile epoxy adhesive (2015 from Huntsman). Table 1 shows the shear

properties of the adhesives used in this work. The properties were determined using

the thick adherend shear test [21].

The heat treated steel DIN 40CrMnMo7 was used for the substrates. It is a high

strength steel with a yield strength of 900 MPa that is sufficient to keep the material in

the elastic range.

2.2 Specimen geometry

The specimen geometry is represented in Figure 2. The geometry is that used for the

double cantilever beam test. Three adhesive thicknesses were studied for each

adhesive: 0.2, 0.5 and 1 mm.

2.3 Specimen manufacture

The joint surfaces were grit blasted and degreased with acetone prior to the

application of the adhesive. Spacers were inserted between the adherends before the

application of the adhesive in order to control the bondline thickness. These spacers

were removed after the adhesive was cured. A sharp pre-crack in the adhesive layer

mid-thickness was assured using a razor blade. A mould with spacers for the correct

alignment of the adherends was used and is shown in Figure 3.

5
2.4 Testing

The ENF specimens were tested in laboratory conditions (~ 25ºC and ~ 50% relative

humidity) using a universal testing machine, under a constant crosshead rate of 0.25

mm/min. The load–displacement (P–δ) curve was registered during the test. Pictures

were recorded during the specimens testing with 5 s intervals using a 10 MPixel

digital camera. This procedure allows measuring the crack length during its growth

and afterwards collecting the P–δ–a parameters. This was performed correlating the

time elapsed since the beginning of each test between the P–δ curve and each picture

(the testing time of each P–δ curve point is obtained accurately with the absolute

displacement and the established loading rate). The specimens were marked with a

white paint and a ruler to facilitate the crack reading.

2.5 Data analysis

According to linear elastic fracture mechanics [22],

Pc2 dC
Gc = (1)
2b da

where C is the compliance defined by C = δ/P, Pc is the load for crack growth and b is

the joint width. According to the beam theory and using Equation (1),

2
9 Pc a 2
GIIc = (2)
16b 2 Eh 3

where GIIc is the toughness in mode I. The toughness in mode II can also be

determined by finding the partial derivative of the compliance with crack length using

an analytical equation that fits the experimental data of the compliance versus the

6
crack length. However, in any case, the experimental measurement of the crack length

is very laborious because the two substrates are against each other and make the

identification of the crack tip very difficult. Also, at the crack tip the fracture process

zone (FPZ) where damage of the material occurs by plasticisation and micro-

crackling is absorbing part of the energy. Therefore, a crack length that takes into

account that FPZ should be used. To overcome these two problems (crack length

measurement and FPZ), de Moura and Morais [23] proposed a method that does not

require the crack length measurement and that takes into account the FPZ that they

called the compliance based beam method (CBBM). The bending modulus can be

obtained considering the initial compliance C0 and the initial crack length a0:

−1
3a 03 + 2 L3 ⎛ 3L ⎞
Ef = ⎜ C0 − ⎟ (3)
12 I ⎝ 10Gbh ⎠

The effect of the FPZ must be included in the compliance and the crack length:

3(a + Δa FPZ ) 3 + 2 L3 3L
C= + (4)
12 E f I 10Gbh

which combined with Equation (3) gives the equivalent crack length:

1
⎡C 2⎛ C ⎞ ⎤ 3
a e = a + Δa FPZ = ⎢ corr a 03 + ⎜⎜ corr − 1⎟⎟ L3 ⎥ (5)
⎢⎣ C 0 corr 3 ⎝ C 0 corr ⎠ ⎥⎦

where Ccorr is given by

3L
C corr = C −
10Gbh

Substituting the value of ae in Equation (2),

2
9P 2 ⎡ C corr 3 2 ⎛ C corr ⎞ ⎤ 3
GIIc = ⎢ a 0 + ⎜⎜ − 1⎟⎟ L3 ⎥ (6)
16b 2 E f h 3 ⎢⎣ C 0 corr 3 ⎝ C 0 corr ⎠ ⎥⎦

7
3. Results

3.1 Brittle adhesive (AV138)

All the specimens failed cohesively in the adhesive, as shown in Figure 4.

Representative experimental P–δ curves of the ENF specimens for each adhesive

thickness are presented in Figure 5. The curves are linear to failure which is in

accordance with the brittle nature of the adhesive. The crack propagation occurred

suddenly after the maximum load. An experimental R-curve obtained for an adhesive

thickness of 0.5 mm is shown in Figure 6. R-curves are used to identify the plateau

corresponding to the crack propagation. A plateau barely appears because the

adhesive is brittle and leads to an unstable crack propagation. Figure 7 shows the

values of GIIc as a function of the adhesive thickness. The brittle adhesive AV138 is

not sensitive to the adhesive thickness and gives an approximately constant value of 5

N/mm. The fracture toughness was determined using the compliance based beam

method (CBBM) because it was not possible to identify the crack tip for crack

measurement due to the sudden and unstable crack propagation.

3.2 Ductile adhesive (2015)

All the specimens failed cohesively in the adhesive, as shown in Figure 8.

Representative experimental P–δ curves of the ENF specimens for each adhesive

thickness are presented in Figure 9. In this case, the curves are non-linear

corresponding to the adhesive plastic deformation. An experimental R-curve obtained

for an adhesive thickness of 0.5 mm is shown in Figure 10. A plateau is clearly seen

indicating a stable crack propagation. Figure 7 shows the values of GIIc as a function

of the adhesive thickness. The fracture toughness in mode II increases almost linearly

8
with the adhesive thickness. The values presented in Figure 7 were obtained using the

CBBM method. However, in the case of the 2015 adhesive, it was possible to measure

the crack length and determine the fracture toughness using the beam theory. Table 2

shows that the beam theory underestimates the GIIc, especially for large bondline

thicknesses (0.5 and 1 mm).

4. Discussion

In adhesive bonding it is important to understand that the adhesive layer applied in

between the two bonded bodies is usually very thin (of the order 0.05 to 0.2 mm), thus

it behaves differently compared to the adhesive as a bulk material. If it is true that

thicker adhesive layers result in bad joint properties, when the adhesive layer becomes

thinner than the surface roughness it is difficult to promote the connection between

the two surfaces because it is difficult to fill the voids. The ability to absorb energy,

characterizing ductile or brittle adhesive plays also an important role when evaluating

the bondline thickness effect. The results presented above show that in the case of a

brittle adhesive, the FPZ is negligible and the adherends do not interfere with the

strain energy release rate measured. However, in the case of the ductile adhesive

2015, the results show that the fracture toughness measured is influenced by the

adherends, as shown schematically in Figure 11. The value of GIIc used for modelling

purposes to design an adhesive joint should be that measured in a fracture mechanics

joint with the same adhesive thickness. This aspect is often not taken into account and

may lead to erroneous results.

The strain energy release rate measured here in mode II can be compared with that

measured in mode I by the same authors in another paper [24]. The fracture toughness

9
in mode I was measured using the double cantilever beam method under a test speed

similar to that used in the present analysis and under the same ambient conditions.

The adhesive thickness that was used is 0.5 mm. The values are presented in Table 3

along with the mode II values and the relation GIIc/GIc. It is common in the literature

to assume a relation of 2 for GIIc/GIc when the value of GIIc is unknown. However, the

results presented here show that the relation can be much higher. Therefore, it is

important to test not only in mode I but also in mode II for the true adhesive

properties. Another study [25] has shown a relation of approximately 10 for GIIc/GIc

for adhesive 2015 but for an adhesive thickness of 0.2 mm. This reinforces the fact

that the adhesive toughness to be used for simulation purposes should use properties

determined in conditions similar to those found in the real structure.

5. Conclusions

The fracture toughness in mode II (GIIc) was measured using the ENF test for a brittle

adhesive (AV138) and a ductile adhesive (2015) using three adhesive thicknesses

(0.2, 0.5 and 1 mm). The following conclusions can be drawn:

1. The critical strain energy release rate (GIIc) for the brittle adhesive AV138 does

not vary with the adhesive thickness and is approximately 5 N/mm.

2. The critical strain energy release rate (GIIc) for the ductile adhesive (2015)

increases almost linearly with the adhesive thickness, varying from 7.15 N/mm for 0.2

mm to 25.8 N/mm for 1 mm.

3. The different behaviour between the two types of adhesives can be explained by

the fracture process zone (FPZ) ahead of the crack tip. In the case of the brittle

10
adhesive, that FPZ is negligible contrarily to the case of the ductile adhesive which

interferes with the adherends.

4. The relation GIIc/GIc for the adhesives studied here is of at least one order of

magnitude.

References

[1] Adams, R. D., Peppiatt, N. A., J. Strain Anal. 9, 185-196 (1974).

[2] Adams, R. D., Harris, J. A., Int. J. Adhes. Adhes. 4, 65-78 (1984).

[3] Hart-Smith, L. J., NASA Contract Report, NASA CR-112236 (1973).

[4] da Silva, L. F. M., Lima, R. F. T., Teixeira, R. M. S., J. Adhesion 85, 889-918

(2009).

[5] Adams, R. D., Mallick, V., J. Adhesion 38, 199-217 (1992).

[6] da Silva, L. F. M., das Neves, P. J. C., Adams, R. D., Spelt, J. K., Int. J. Adhes.

Adhes. 29, 319-330 (2009).

[7] da Silva, L. F. M., das Neves, P. J. C., Adams, R. D., Wang, A., Spelt J. K., Int. J.

Adhesion Adhesives 29, 331-341 (2009).

[8] Xu, J.-Q., Liu, Y.-H., Wang, X.-G., Engineering Fracture Mechanics 63 (6), 775-

790 (1999).

[9] Groth, H. L., Int. J. Adhes. Adhes. 8, 107-113 (1988).

[10] Gleich, D. M., Van Tooren, M. J. L., Beukers, A., J. Adhesion Sci. Technol. 15 (9),

1091-1101 (2001).

[11] Duan, K., Hu, X., Mai, Y.-W., J. Adhesion Sci. Technol. 18(1), 39-54 (2004).

[12] Needleman, A., J. Applied Mechanics 54, 525-531 (1987).

[13] Ungsuwarungsri, T., Knauss, W. G., Int. J. Fracture 35, 221-241 (1987).

11
[14] Tvergaard, V., Hutchinson, J. W., J. Mechanics and Physics of Solids 40, 1377-

1397 (1992).

[15] Bascom, W. D., Cottington, R. L., Jones, R. L., Peyser, P., J. Appl. Polym. Sci.

19, 2545–2562 (1975).

[16] Bascom, W. D., Cottington, R. L., J. Adhesion 7, 333–346 (1976).

[17] Kinloch, A. J., Shaw, S. J., J. Adhesion 12, 59–77 (1981).

[18] Daghyani, H. R., Ye, L., Mai, Y. W., J. Adhesion 53, 149–162 (1995).

[19] Daghyani, H. R., Ye, L., Mai, Y. W., J. Adhesion 53, 163–172 (1995).

[20] Lee, D., Ikeda, T., Miyazaki, N., Choi, N., J. Engineering Materials and

Technology 126, 14-18 (2004).

[21] Marques, E. A. S, da Silva, L. F. M., J. Adhesion 84, 917–936 (2008).

[22] Irwin, G. R., Appl. Mater. Res. 3, 65-81(1963)

[23] de Moura, M. F. S. F., de Morais, A. B., Eng. Fract. Mech. 75, 2584–2596

(2008).

[24] da Silva, L. F. M., Carbas, R. J. C., Critchlow, G. W., Figueiredo, M. A. V.,

Brown, K., Int. J. Adhes. Adhes. 29: 621–632, 2009.

[25] Campilho, R. D. S. G., de Moura, M. F. S. F., Ramantani, D. A., Morais, J. J. L.,

Domingues, J. J. M. S., Int. J. Adhes. Adhes. 29, 678–686 (2009).

12
Table 1 Adhesive shear properties using the thick adherend shear test method ISO 11003-2
[11].
AV138M / HV998 2015
Shear modulus
1559 ± 11 487 ± 77
G (MPa)
Shear yield strength
25.0 ± 0.55 17.9 ± 1.80
τya (MPa)
Shear strength
30.2 ± 0.40 17.9 ± 1.80
τr (MPa)
Shear failure strain
5.50 ± 0.44 43.9 ± 3.40
γf (%)

Table 2 Fracture toughness in mode II (GIIc) determined using the beam theory and the
CBBM method.
Adhesive thickness (mm) Beam theory GIIc (N/mm) CBBM GIIc (N/mm)

0.5 11.3 13.2

1 21.2 32.4

Table 3 Comparison of the fracture toughness in mode I (GIc) and mode II (GIIc) for an
adhesive thickness of 0.5 mm.
Adhesive GIc (N/mm) [Carbas] GIIc (N/mm) (present study) GIIc / GIc

Brittle (AV138) 0.346 4.91 14.2

Ductile (2015) 0.526 11.9 22.6

13
 y

Figure 1 Schematic representation of the end notched flexure (ENF) test, end loaded
split (ELS) test and four-point notched flexure (4ENF) test methods.

25.4
1 2 .7
6 .3 5

6.35
a 0 = 51
356

Figure 2 Dimensions in mm of the end notched flexure (ENF) test specimen.

14
Figure 3 End notched flexure specimen fabrication (shims for bondline thickness
control at the top and assembled specimens in a mould at the bottom).

15
Figure 4 Failure surfaces of end notched flexure (ENF) specimens with the brittle
adhesive AV138.

25000

20000 0.2 mm

15000
P (N)

10000 1 mm

5000
0.5 mm
0
0 0.5 1 1.5 2
δ (mm)

Figure 5 Representative experimental P–δ curves of the ENF specimens with the
brittle adhesive AV138 as a function of the adhesive thickness.

16
 6

5.5
GIIc (N/mm)

4.5

4
55 60 65 70
aeq (mm)

Figure 6 Typical experimental R-curve obtained for the brittle adhesive AV138 for a
thickness of 0.5 mm.

35

30 2015
AV138
25
GIIc (N/mm)

20

15

10

0
0 0.2 0.4 0.6 0.8 1 1.2

Adhesive thickness (mm)

Figure 7 Mode II fracture toughness (GIIc) as a function of the adhesive thickness for a
ductile adhesive (2015) and a brittle adhesive (AV138).

17
 Figure 8 Failure surfaces of ENF specimens with the ductile adhesive 2015.

Figure 9 Representative experimental P–δ curves of the ENF specimens with the
ductile adhesive 2015 as a function of the adhesive thickness.

18
 14
13
12
11
GIIc [N/mm]

10
9
8
7
6
5
4
3
55 60 65 70 75 80 85
aeq [mm]

Figure 10 Typical experimental R-curve obtained for the ductile adhesive 2015 for a
thickness of 0.5 mm.

Figure 11 Fracture process zone (FPZ) as a function of the adhesive bondline thickness.

19