S Mehdi Tavakoli, TWI, Cambridge, UK
Ewen J C Kellar, TWI, Cambridge, UK
Cosmas Vlattas, European Marine Contractors Ltd, New Malden, UK
Paper 402 presented at Society of Plastics Engineers Annual Technical Meeting, SPE ANTEC 2001 Conference - Medical Plastics, 6-10 May 2001, Dallas, Texas, USA.
Single lap shear specimens were prepared using a zinc coated steel adherend and an epoxy adhesive and evaluated by static and fatigue tests. Joints fatigue tested in air failed in a cohesive manner with some zinc delamination at the highest load. Joints tested in water failed through a combination of adhesive and cohesive failure. SEM, EDX, and FTIR analyses enabled a detailed characterisation of all the components of the epoxy system. Despite a generally homogenous distribution within the bulk of the adhesive, an interfacial layer, devoid of particulates has been identified next to the zinc layer.
The use of adhesives in a range of industrial applications (e.g. electronics, automotive, aerospace and medical) has increased significantly in recent years. However, a recent survey has identified that prediction of lifetime of bonded joints is one of the most significant factors restricting the use of adhesive technology  . Despite significant research into durability, there remains considerable uncertainty in predicting joint lifetime.
Work has been carried out at TWI to study durability and to develop methods for investigating the life expectancy of bonded joints [2-5] . The objectives of this project were to acquire fatigue endurance and crack growth data from a galvanneal steel adherend/epoxy adhesive system and to correlate the structural performance of the joints with key constituents and fracture appearance.
- Zinc coated steel - a galvanneal steel (Galvatite ZF (12)): sheet thickness 1.25mm; Zn layer 10-15µm.
- Stainless steel.
A one-part automotive grade (XB5315) heat-curing epoxy. XB5315 is based on Bisphenol-A epoxy (32-44%) resin and contains various organic (rubber) and inorganic mineral fillers and dicyandiamide (the hardener component).
Single Lap Shear (SLS) Specimens
SLS specimens are shown in Fig.1
. The SLS specimens were prepared either individually or cut from a panel in groups of at least five.
Fig.1. Single lap joint geometry and dimensions
Where specimens were prepared individually, the adhesive was applied to the area across the end of one of the metal substrates. Two 0.3mm thick steel wires were carefully located in the bond area across the overlap to control the bondline thickness. These wires were permanent features in the joint after curing. Using a simple clamping arrangement, the joints were held rigidly during curing (180°C for 30 minutes in a fan assisted oven).
Alternatively, the specimens were made up by bonding sheets of metal, to produce a panel from which specimens were cut. The wires controlling the bondline thickness were inserted longitudinally to the joint axis orientation and discarded at the cutting stage. Thus, there were no wire spacers in the final specimens.
Single Lap Shear Specimens
Three specimens per environmental condition (air RT, air 60°C and water 60°C) and type of joint (with or without wire spacers) were tested statically (crosshead speed - 10mm/min). The peak failure load was recorded. Fatigue tests were carried out in similar environmental conditions and configurations (frequency - 10Hz; stress ratio - 0.1). The stress range and number of cycles to failure were recorded.
The fatigue performance of SLS joints was measured at room temperature and 60°C.
For the tests at 60°C, an environmental chamber was used. The specimen was allowed to equilibrate for 30 minutes prior to testing.
Fatigue tests were also conducted at 60°C, with specimens fully immersed in water.
Optical and scanning electron microscopy (SEM) (CamScan CS44 with PGT/EDX) as well as fourier transform infrared microscopy (FTIR, Mattson Polaris) were used.
Results and discussion
As can be seen in Fig.2
, the existence of the wire spacers does not significantly affect the static strength of the joints. However, the effect of the temperature is clear: the static shear strength of the joints tested in air at 60oC was reduced by 30%, compared to the static strength of the joints tested in air at room temperature. No further decrease of the static strength was observed for joints tested in water at 60oC.
Fig.2. Static strength data for single lap shear joints
Fatigue data were generated by subjecting single lap shear specimens to a wide range of constant load amplitudes. The results for all specimens are plotted in logS-logN diagrams in Fig.3. As can be seen, the presence of wires controlling bondline thickness significantly affected fatigue performance. Linear regression curves in log-log scale were fitted to the raw data. The fatigue strength defined at 10 7 cycles was calculated using these regression curves. In Table 1, the fatigue strength data and static data are summarised. The ratio of the fatigue strength to the static strength is also tabulated. This is an indicative parameter of environmental degradation of the strength of the joint. The fatigue results for all environmental conditions are presented in Fig.4.
Fig.3. S-N curves for galvanneal steel/XB5315 single lap shear joints Fig.3 a) with wire spacer
Fig.3b) without wire spacer
Fig.4. Fatigue strength date for single lap shear joints
Table 1 S-N characteristic parameters for single lap joints
|Condition||Static Shear Strength, MPa||Fatigue Shear Strength*, MPa||Ratio|
|Samples with wire spacer
|Samples without wire spacer
|* at 10 7 cycles.
The fatigue shear strength of joints tested in air at 60°C is reduced by 50% for joints with wire spacers and 35% for joints without wire spacers, compared with results in air at room temperature. A further decrease in the fatigue shear strength was observed for those tested in water at 60°C (85% for the joints with the wire spacers and 75% for the joints without the wire spacers). Clearly, the reduction in fatigue shear strength is enhanced by the presence of wire spacers.
Adhesive composition and interaction with the zinc coating
A secondary electron SEM image of a mixed adherend reference joint is shown on the left hand side of Fig.5. For the galvanneal steel adherend, the zinc coating is variable in thickness (10-15µm) containing cracks and voids, characteristic of such coatings. The adhesive layer contains a high volume of filler particles of varied size and shape (<1µm to >300µm). Distribution of these particles is homogeneous, except for a layer adjacent to the zinc coating. This layer, 15-20µm thick, is almost totally devoid of any particulate material, in contrast to the stainless steel surface that shows no adjacent interfacial layer.
Fig.5. Combined secondary and backscattered electron SEM image of sectioned adhesive layer.
A - filler material (talc - irregular, wollastonite - needle-like)
B - epoxy resin
C - rubber regions
D - interfacial layer
E - zinc coating
F - mild steel substrate
The backscattered SEM image shown on the right hand side of Fig.5, reveals a high level of contrast between the dark background of the bulk adhesive resin and the light grey inorganic particulates within it. The interfacial layer above the zinc coating can also be identified, being considerably darker than the bulk resin material.
Spot EDX analysis was carried out on selected areas as indicated in Fig.5. The results are summarised as follows:
- High levels of calcium and silicon with some magnesium, indicating that they are either talc, (irregular) or wollastonite (needle-like).
- The light grey regions within the bulk material contain carbon, silicon and oxygen and are due to epoxy resin containing silica and perhaps some silane material.
- Dark grey regions within the bulk material contain more carbon and smaller amounts of silicon and calcium. It is suggested that these areas are composed of rubber toughening material.
- The interfacial layer above the zinc coating contains high levels of carbon with trace levels of calcium, and silicon.
The area described was also EDX mapped to show the overall distribution of selected elements, Fig.6. Of greatest interest is the contrast between the interfacial layer and the bulk resin, with the former showing more carbon, slightly less oxygen and very low levels of silicon. This implies that the interfacial layer has a greater hydrocarbon concentration that the bulk resin and that silicon, in the form of silica is present only at very low levels.
Fig.6. EDX element maps of sectioned adhesive layer
The presence of little or no inorganic filler within the layer is not easily explained, especially as all sizes of particulate material are absent. It is assumed that electrostatic repulsion, combined with some type of phase separation may play a role, but no work to date has been carried out to verify this.
It should be noted that, although the layer can be clearly observed for some sample sections, it is not always evident and sometimes present only as globular patches above the zinc coating.
Joint fracture surfaces
(a) Air at room temperature
Light microscopy revealed that the samples with wires (static and fatigue loading) failed in a predominantly cohesive manner with some zinc delamination for static loading or at high fatigue load ranges. Adhesive failure was also seen next to the wire spacers. No trends relating to loading and mode of fracture were identified.
Analysis of the samples without wires revealed that there was limited zinc delamination present for some joints. The primary failure mode appeared to be cohesive in type.
(b) Air at 60°C
Cohesive failure was the main failure mode for all samples tested in air at 60°C. As for the room temperature samples, the presence of wire spacers within the bondline ('old' specimens) would appear to trigger failure as evidenced by adhesive failure around the wires.
(c) Water at 60°C
The mode of failure for joints with wire spacers loaded in fatigue was a combination of cohesive and adhesive failure. The latter often covered up to 40% of the joint area and would appear to be related directly to the presence of the spacer wires.
The presence of wires within the bondline was observed to be critical in determining the mode and locus of failure for both the static and fatigue tests. The interface between the adhesive and the wires enabled water to penetrate the joint and attack the interface giving rise to adhesive failure.
Similarly, the joints without wires failed by a mixture of cohesive and adhesive failure. However, the adhesive failure pattern followed a distinct trend, in that it was crescent shaped and focused at each end of the joint. The pattern increased in size with decreasing loads. This corresponded to increasing exposure times and is related to the rate of diffusion of water through the adhesive.
FTIR analysis was carried out on four samples, selected to cover the greatest range of test conditions.
Variations in the spectra were observed but there were few consistent changes which could be ascribed to changes in the chemical structure of the material. These variations were greatest between 1000 and 1100cm -1 where two or three very strong peaks varied in intensity. This area corresponds to vibrational bands of the inorganic filler reinforcement material. It is assumed that such variations are purely due to changes in the distribution of the filler from sample to sample.
Figure 7 shows a single FTIR spectrum with appropriate assignments.
Fig.7. Transmission infrared spectrum of cured XB5315 adhesive, with main group assignments
The main conclusions were as follows:
- The inclusion of permanent spacer wires within the bondline of an adhesively bonded joint has been shown to significantly compromise its fatigue strength and its fatigue life. It is therefore strongly recommended that wherever possible, this method of bondline thickness control is not used. The presence of wires creates defects or cracks within the joint from which failure can initiate, either through cohesive failure within the adhesive or by hydrological attack at the adherend/adhesive interface.
- All single lap shear joints fatigue tested in air failed cohesively with some zinc delamination evident at the highest loads. Joints tested in water failed through a combination of adhesive and cohesive failure. Adhesive failure was attributed to moisture diffusion into the joint and a clear trend was seen in terms of increasing levels of adhesive failure with length of exposure when no wire was present.
- A reaction between the adhesive and the zinc coating to form an interfacial layer has been shown to occur for some bonded specimens. However, no evidence has been found to link the presence of this layer with the mechanical performance of the adhesive joint in either static or fatigue loading conditions.
- McGrath G C, Int. J Adhesion and Adhesives 17 339-342 (1997)
- Tavakoli S M, TWI Members Report 548/1996, May 1996.
- Kellar E J C and Tavakoli S M, Adhesion '99, 7 th International Conference on Adhesion and Adhesives419-424 (1999).
- Kellar E J C, Vlattas C and Tavakoli S M, Adhesion '99, 7 th International Conference on Adhesion and Adhesives 295-300 (1999).
- Court R S, Sutcliffe M P F and Tavakoli S M, Adhesion '99, 7 th International Conference on Adhesion and Adhesives 237-242 (1999).