Sei sulla pagina 1di 7

Adhesive bonding of fibrereinforced composites

B.M. Parker
(Defence Research Agency, UK)

The applicability of test methods for adhesive-bonded metal adherends to fibrereinforced composites is considered. Bond strengths obtained by methods chosen as suitable are compared. The most critical factor in metal joints is the pretreatment of the adherend before bonding, as this controls the structure and chemistry of the oxide surface and hence durability in hot-humid conditions, often with little effect on initial strengths. For fibrereinforced epoxy resin adherends, the critical factor is the initial bond. Initial bond strength, measured by any method, is related to the presence of contaminants on the adherend surface. Loss of strength in hot-humid conditions is related to water absorption by adhesive and matrix and not to reduced adhesion. Thermoplastic-matrix composites require pretreatments which increase their surface energy and wettability by adhesives.

Key words: adhesive-bonded joints; composite adherends; durability; test methods; surface contamination

Any structure, whether it is a motor car, an aeroplane or a lawn mower, needs to be suitable for the job that it is expected to do and to last. That is, it has to be strong enough initially to meet some standard and it must retain its strength for its expected lifetime; it must be durable. An airframe is first and foremost exposed to vibration and the stresses of take-off and landing which result in fatigue damage. This results in cracking and is well understood for metals. Corrosion of metals is also a problem. When organic materials such as fibre-reinforced composites and adhesives are introduced into structures, a further problem is also introduced, that of the effect of absorbed atmospheric moisture on the organic material. In composites, the resin-matrix dominated properties are reduced. This also applies to the properties of adhesives. But of much greater concern in the past has been the effect of moisture on the adhesive-to-metal bond. This problem has long been recognized and has resulted, in the aircraft industry, in an enormous volume of work on the surface pretreatment of metals, particularly aluminium,

before adhesive bonding. With the service lives of aircraft being extended beyond what was probably initially envisaged, 'ageing aircraft' are now a major concern. Fibre-reinforced composites are being more and more widely used in civil as well as military airframes and it is necessary to know what their durability is likely to be and whether any special measures are required to improve it. This paper considers the applicability to composites of methods which are used to test the strength of bonded metal joints, using carbon fibre-reinforced composite; tests are then compared to see whether they give the same information on bonding. Lap joints and wedge specimens were chosen for a study of the effect of absorbed moisture on joint strength. The results obtained are compared.

Test methods for composite joints


The main tests used for metal-to-metal joints are the lap shear joint (single or double), tensile butt joint, floating roller peel, T-peel and wedge test. All have

0143-7496/94/02/0137-07 ~;) 1994 Butterworth-Heinemann Ltd ~'~Crown Copyright, 1993 INT.J.ADHESION AND ADHESIVES VOL. 14 NO. 2 APRIL 1994 137

been used to assess durability by exposure to hot humid conditions, and some have also been subjected to natural, static, weathering. Lap shear joints There are no problems in translating single or double lap joints from metal to composite. The dimensions shown in Fig. 1 are those recommended by CRAG, Method 102 (specimen A) i. Floating roller peel This specimen (Fig. 2) (EN 2243 Part 02, similar to ASTM D3167) consists of a thick base member and a thin member, which has to be flexible enough to bend round the roller. The same restriction on the thin member applies as for T-peel. This test can be used to test the bondability of unaged composite by using a thin metal member and a thick composite base. However, the use of a thin metal member would cause complications in durability studies due to the effect of moisture on the adhesive/metal interface. Unless it is actually intended to study metal-to-composite joints, this specimen is not suitable for durability studies. Wedge test This test, to ASTM D3762 for metals, has been adapted for composites as shown in Fig. 3. The length of 100ram was the result of using this as the standard

panel length for lap joints; a longer specimen is needed if it is really necessary to know crack lengths liar "bad' ,joints. Specimen edges arc polished to at least 400 grit. T-peel This is shown in Fig. 4. There are problems in adapting this for carbon fibre-reinforced composite due to anisotropy of the base material. As with the floating roller peel, a very thin laminate is required to be able to bend the unbonded ends enough to grip them. This is a minimum of a 4-ply laminate lbr unidirectional material, or two layers of cloth. It is possible that a (45)~ laminate could be used to reduce modulus in the direction of the test. A 2-ply cloth laminate has been used in surface studies:, but, if the bond is good, there is a problem with failure of the adherend across the end of the bondline. Tensile butt joint The metal specimen uses bar stock (Fig. 5). There are practical difficulties in adapting this for composites. If a bar is machined with the bondline in the same plane as the fibre at the composite surface, testing in tension would be testing transverse to the fibre plane and the composite strength would be dependent on the matrix, which is usually more brittle than most structural adhesives. There is probably no reason why this specimen could not be used for brittle adhesives. Comparison of test methods on unaged joints Experimental All the specimens described in this part of the work were made from unidirectional Ciba-Geigy Fibredux XAS/914C epoxy-matrix composite, nominally 2 m m thick. The surface was varied by curing against a range of release agents 3, which left polytetrafluoroethylene (PTFE), silicone or polyamide residues on the composite

~_ ~ h e s i v e

filet

Fig. 1 Single lap shear composite test specimen, all dimensions in mm

Roller earing

Steel rollers

;i
l

"/
5~
1 25

,i

30

i"

I'i I
25 _

I
4

1
Fig. 2 C o m p o s i t e peel test after EN2243-2 m e t a l - t o - m e t a l peel test, all d i m e n s i o n s in m m

138

INT.J.ADHESION AND ADHESIVES APRIL 1994

I_

I
100

~Adhesive
r ~

25

-T 25
-y
Adhesive

25

cr ck teo h
Fig. 3

Composite wedge test, all dimensions in mm


280 mm ._ 0.37 mm

i
,

.,

Fig, 5 Tensile butt joint, all dimensions in mm

Joint strengths

Fig. 4

Composite T-peel specimen

surface. The level of contamination was reduced either by hand abrasion with an abrasive cloth or by blasting with dry alumina grit. The nature and level of contamination was estimated by X-ray photoelectron spectroscopy (xes) which showed 3 that surface contamination is partly removed by hand abrasion with carborundum and, except in cases of very heavy contamination, virtually completely removed by dry grit-blasting. All surfaces were wiped clean of loose debris with an acetone-soaked tissue before bonding. The adhesives used were all modified structural epoxies. E was a twopart paste cured for 1 h at 60C, B and J were film adhesives cured for 1 h at 120C, and P and R were films cured for 1 h at 175C. L, cured for 1 h at 175C, was apparently untoughened and far more brittle than all the others. Wedge joints were made individually from pieces of composite 100mm long by 25mm wide; using a standard 25 mm long wedge, this limited the maximum crack length to about 80 mm. Wedges were inserted and crack lengths measured using a travelling microscope before and after exposure to 50C/96% relative humidity (RH) for up to 900 h. Joints were then broken open. Single overlap joints were made with a 12.5 x 25 mm overlap. Joint strengths in tension were measured before and after exposure to 50C/96% RH for 6 weeks. The mode of failure of broken joints was estimated visually. For wedge joints, the initial failure mode was examined. The three modes that could be differentiated were interracial failure between adhesive and composite, delamination of the composite and cohesive failure in the adhesive layer. Frequently, all three were present in a single joint.

Previous work has shown a correlation between the level of contamination on the composite surface, the unaged strength of adhesive-bonded joints 3 and initial crack length*, and between floating roller peel strength and lap joint strength. A series of tests was performed in which lap shear, floating roller peel and wedge joints were all made from the same laminate panel, so that the extent of surface contamination transferred from the mould release agent to the laminate surface could be assumed to be about the same in each specimen. All these comparisons were made using adhesive E. Strength data for unaged lap shear and wedge joints are given in Table 1, together with the major mode of failure and its estimated percentage. As can be seen, composites which gave interfacial failure in lap joints gave the same failure in wedge joints, and composites giving failure in the adhesive gave the same in wedge joints. However, composites for which the lap joints failed in the composite gave wedge joints which failed in the adhesive. This was also the case for peel joints. It is therefore legitimate to compare the different types of joint. The results for lap shear strength are compared with those for peel strength in Fig. 6 and with initial wedge crack length in Fig. 7. It is clear from Fig. 6 that joint strength and floating roller peel strength (note that the peel strength scale is logarithmic) are related to each other. As increasing severity of treatment reduces surface contamination, the strengths of both joints increase and the mode of failure changes from being at the interface between composite and adhesive to failure away from the interface. The data in Fig. 7 tell much the same story: crack length decreases as joint strength increases and the mode of failure changes. However, in this case, joint strength appears to have reached a maximum for crack lengths below about 55 mm, which can be achieved with only partial removal of contamination. Note that, because of the specimen dimensions, the maximum crack length was 80 mm, at which the specimen fell

INT.J.ADHESION AND ADHESIVES APRIL 1994

139

5O Joint strength (MPa) 40

50 Joint strength (MPa) 40

1 1

i
8.,,b o
()

cP
30 O Failure at composite interface 20


I I 1 I
[
@

]
@

O O

1 t I
In

3O

]Mixedl Fa.ilure [adhesive

OoO
O O 10 _ 0 O
0 I I I I

2O O Untreated Hand abraded Grit blasted O O

O
O O 0 O 0 O

o0 o
O Untreated Hand abraded Grit blasted
I I I

10

Failure in adhesive or composite

Failure at interface

0.1 0.2
Fig. 6

0.5 1.0 2.0 5.0 Peel strength (kS/m)

10

20

I
80

30

40

Comparison of lap joint strength and peel strength

50 60 70 Initial crack length (mm)

apart. Use of longer adherends could result in a range of crack lengths > 80 mm for the points shown at this length.

Fig. 7 Comparison of lap joint strength and wedge test initial crack length

Testing joint durability


For the reasons given above, only lap and wedge joints seem to be suitable for studying the durability of composite-to-composite joints. Both have been used widely for durability studies on metal-to-metal joints. Recent analysis of outdoor weathering data on lap joints and laboratory exposure of wedge specimens shows that a correlation can be made between them s.

strength of joints of L was half that of the other adhesives. The absorption of water during exposure was measured by weighing the joints before exposure and before testing. The average strength loss, relative to dry

Lap joints
The effect of prolonged exposure to hot-humid conditions has been studied for single lap composite joints 6, v. Joints made with grit-blasted adherends were exposed unstressed to 50C/96% RH for up to 3 years. Residual strength was measured as a function of temperature. Different adhesives give joints with different initial strengths, so results have been normalized by taking all residual strengths as a percentage of initial room temperature strength. The effect of absorbed moisture as a function of temperature is shown in Fig. 8; open symbols are unaged joints and closed symbols aged ones. With the exception of unaged joints of the brittle adhesive L, strength decreased with increasing test temperature. Hot-wet exposure generally reduced joint strength. After ageing, joint retentions followed approximately parallel curves with the 120C-cured adhesive showing lower retention than two modified 175C-cured ones, and all showing much poorer retention than adhesive L. However, it must be noted here that the initial

140 Retention of dry 20C I20 strength

(%)
100

80 60 40 20 v~ \

210

410

60

810

100

120

Test temperature (C)


Fig. 8 Summary of test temperature and exposure on joint strength reduction for joints bonded with adhesives B (O, I l L J (~), L ([C], II), P (A, A) and R (V, V). Open symbols, unaged joints; closed symbols, joints aged for 25 000 h

140

INT.J.ADHESION AND ADHESIVES APRIL 1994

joints tested at the same temperature, was calculated for each test temperature and exposure time. Strength decrements for the brittle adhesive L were temperaturedependent but appeared to be temperature-independent for other, modified or toughened adhesives; for these, an average decrement was calculated over all test temperatures. The average losses in strength are shown as a function of water content in Fig. 9. At low water contents, average losses lay within a narrow band for all adhesives, but differences between adhesives began to appear at > 2% water content. This may indicate that, at short times/low water content, changes in joint strength are controlled by the laminate, but that at longer times the adhesive becomes the controlling factor. Joints were also made with untreated adherends, but were only aged for 6 weeks. The results, expressed in terms of percentage strength retention, are given in Table 1 together with the major modes of failure. Untreated joints are not specifically identified but are generally the ones with low initial strengths which failed along the interface. As can be seen from Table 1 only in one case was there a change in the mode of failure, and it was exceptional for there to be an appreciable change in the proportion of that mode.
Wedge joints

20 Loss in strength 15 (MPa) 10

zx/ /

-5

10

0.5

1 1 210 2.5 Joint water content (%)

'o

'5

3.0

Fig. 9 Effect of joint water content on loss of joint strength: O, adhesive B, all temperatures; [3, adhesive P, all temperatures; x, adhesive R, all temperatures; 0 , / % A , adhesive L at 120C, 80C and 20C + 60C, respectively

The wedge test was developed specifically to study the durability of metal adherend j o i n t s8 , by exposing stressed specimens to a hot-humid environment and following crack growth. Composite wedge specimens were also exposed to 50C/96% RH and crack growth followed for up to 168 h. The 48 h crack length ratios are given in Table 1. The change in crack length over 48 h for a selection of specimens is shown in Fig. 10. The time is fairly arbitrary but experience has shown that most crack

growth takes place within this time. Only contamination type and mode of failure are identified. From this it appears that, independent of contamination and initial crack length, crack growth usually was greater for specimens which failed at the interface than for those which failed in the adhesive or adherend. Crack growth also tended to be greater for specimens with longer initial cracks. From closer consideration of the nature of the surface contamination for a wide range of specimens, it can be concluded that crack growth is generally low

Table 1. Joint strength retention after 1000 h and crack growth after 48 h at 50C/96% RH Lap joint
Initial strength (MPa) 9.4 28.6 30.1 35.2 31.7 22.4 31.8 41.1 31.6 34.3 40.2 47.0 34.8 35.7 35.5 38.4 9.1 Failure mode (%)* 95 89 93 76 64 I I A A A Retention (%) 146 130 104 104 103 103 102 102 98 97 96 91 91 87 84 79 74 Exposed Initial crack length (mm) max 51.1 36.7 28.0 31.6 47.2 60.5 34.2 32.4 36.0 39.8 39.5 64.0 42.6 65.7 Wedge test

Failure
mode (%)* 100 I 97 I 50 A 79 A 70 A 90 I 99 I 82 61 98 63 68 A A A A A

ao/ a 48

Failure
mode (%)* 99 93 96 70 80 I I A C C

0.873 0.977 0.836 0.927 0.957

90 I 86 A 91 C 94 A 90 A 85 A 89 C 82 A 68 I 80 A 56 A 100 I

87 I 97 A 55 C 96 A 91 A 98 A 89 C 95 A 62 I 89 A 64 C 100 I

0.909
0.988 0.970 0.947 0.974 0.981 0.989 0.968 0.881

100 I 70 A 100 I

*1, failure at interface; C, failure in composite; A, failure in adhesive

INT.J.ADHESlON AND ADHESIVES APRIL 1994

141

8O Crack length mm 70

U,I --Maximum-

140 Joint strength retention 120 % 100


80

Ox
X

60
600.80

~o
50

0'.90 Crack growth, ao/a48

1.00

Fig. 11 Comparison of measures of durability: O, failure at interface; O, failure in adhesive; x, failure in composite

0--0----~
40

I _/_~~
i ~"

30

outcome of the effect of water on the epoxy resins if plasticization reduces the strength of the adhesive and also toughens it. The fact that a wedge test crack grows, therefore, does not give reliable information on what might happen to lap joint strength, which is stressed in a different way, on exposure.
Thermoplastic-matrix composites

20

48

Exposure time, hours (arbitrary scale)


Fig. 10 Crack growth as a function of surface contamination (O, PTFE; *, silicone; O, polyamide) and failure mode (I, failure between adhesive and composite; A, failure within adhesive; C, failure within composite)

and is only appreciable when the composite surface is contaminated with PTFE or silicone, but not with polyamide, and the joint fails along the adhesive/ composite interface. A durable joint in a metal wedge test is one where the crack growth is low, and failures do not run along the metal/adhesive interface. In these terms, composite joints which fail along the interface initially due to contamination are likely to be less durable. In a durable joint, any crack growth that takes place results from the effect of absorbed water on the adhesive. Crack tips are often visibly blunted as the adhesive yields. By analogy with strength retention of lap joints, wedge tests can be assessed by taking exposed crack length compared with initial crack length. Since crack length increases with time, it is convenient to divide initial crack length, a0, by crack length at the chosen time so that the smaller the ratio, the less durable the joint. The ao/a4~ ratios are given in Table 1.
Comparison of durability tests

In all the work described above, epoxy-matrix composite was bonded with an epoxy adhesive. Cured epoxy resins would be expected to be easily wetted by epoxy resins as they should be compatible chemically and thermodynamically. However, most thermoplastics have low surface energies and are not readily wetted by adhesives. Thermoplastics such as polyethylene and polypropylene require pretreatment before bonding in order to raise their surface energies and make them wettable by adhesives. Considerable effort has been put into the pretreatment of the engineering thermoplastic PEEK (polyetheretherketone), used in combination with carbon fibres as APC-2 composite 9, with corona discharge pretreatment being the current favourite.
Conclusions

The 'durability factor' data for both types of exposed joint are plotted in Fig. 11. Given the scatter, it is difficult to reach any conclusion as to whether the tests correlate. If the two points at 74% and 146% retention are ignored, it might be concluded that joint strength retention is decreased only when the crack growth ratio is approaching 1.00, that is when very little crack growth is taking place. This could be the logical

The results of testing single lap, floating roller peel and wedge joints made with epoxy composite with different amounts and types of surface contamination show that a correlation exists between the different tests. However, exposure tests for durability in hot humid conditions add nothing to the argument: if the surface has already been shown to be bad by the initial tests, joint strength is not generally further reduced, and long cracks grow longer. If joint strengths are better and cracks short, changes on durability testing are due to changes in the adhesive and not to changes in the bond between adhesive and composite. Therefore no additional information is gained from durability testing, and initial strength or crack length data and mode of failure are sufficient to tell whether a surface is suitable for b o n d i n g . These results are specific to epoxy-matrix composites bonded with epoxy adhesives. The test methods should apply equally to thermoplastic-matrix composites, but it should be noted that they do require surface pretreatment to increase their wettability by adhesives.

142

INT.J.ADHESION AND ADHESIVES APRIL 1994

Acknowledgements
This work has been carried out with the support of Procurement Executive, Ministry of Defence.

5 6 7 8 9

References
1 2 3 4 Curtis, P.T. (Ed) Technical Report TR 88012 (Royal Aerospace Establishment, Farnborough, 1988) Hamer, J. BAe Report AL/MAT/3985 (British Aerospace, 1986) Parker, B.M. and Waghorne, R.M. Surf Interface Anal 17 (1991) p 471 Parker, B.M. "Bonding and Repair of Composites" (Butterworths, London, 1989) p 51

Parker, B.M. unpublished results Parker, B.M. Technical Report TR 89001 (Royal Aerospace Establishment, Farnborough, 1989) Parker, B.M. Int J Adhesion and Adhesives 10 (1990) p 187 Marceau, J.A., Moji, Y. and McMillan, J.C. Adhesives Age 20 (1987) p 28 Kodokian, G.K.A. and Kinloch, A.J. J Mater Sci Lett 7 (1988) p 625

Author

The author is with the Materials and Structures Department, Defence Research Agency, Farnborough, Hants GU14 6TD, UK.

I N T . J . A D H E S l O N A N D A D H E S I V E S A P R I L 1994

143

Potrebbero piacerti anche