Adhesive Imperfections and Fracture Behavior

Figure 14 presents an obviously perfect behavior of point supports under tensile loading for a monotonic load history. Despite the special test fittings allowing an almost perfect bonding geometry and despite careful application of the adhe­sive to the specimens, the mechanical characteristics of the tested point sup­ports show a significant scattering, as can be seen in Fig. 19.

The ultimate failure of the specimens occurs at different strain and load lev­els. Furthermore, samples 5 and 3 show a degradation of the mechanical charac­teristics significantly before the final failure. This poses the question of why the investigated specimens differ in their mechanical characteristics although manu­facturing was carefully done under laboratory conditions. In order to identify indications for the different failure behavior, the fracture surfaces of the circular bonding specimens were investigated. Figure 20 displays photographs of the frac­ture surfaces of the specimens corresponding to the load curves shown in Fig. 19.

The following statements can be drawn by studying these fracture surfaces in detail:

• Only two specimens do not show any flaws and can thus be considered perfect; these are specimen 1 and specimen 8. The test of specimen 8 was stopped bevore total break. Afterwards the specimen was cut up in two pieces for analysis of the core material.

FIG. 19—Load curves for point supports with and without adhesive defects.

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014 Downloaded/printed by

FIG. 20—Characteristic fracture surfaces of specimens.

• Specimen 4 shows a flaw near the center of the circular bonding, i. e., within an inner circle of 30% of the total test specimen radius.

• Specimens 2, 3, and 5 show flaws in offset positions with respect to the centre, i. e., in an annulus ranging from about 30 to 60% of the total test specimen radius.

• Specimen 1 and 8 present a fracture surface of high regularity, i. e., a “rose" pattern almost perfectly centered in the circular bonding.

• This rose pattern is disturbed in the other specimens by the flaws. The severity of the disturbance of this pattern is as follows: specimen 4 > specimen 2 > specimen 5 > specimen 3.

As a next step, the establishment of a relationship between the fracture sur­face pattern and the mechanical fracture behavior was considered. According to Fig. 19, specimens 1, 4, and 8 show a high performance in terms of successfully resisting high loads and/or high strains. This high performance can obviously be attributed to the total absence of flaws or to flaws located near the center of the bonded specimen. On the contrary, the performance of specimens 2, 5, and 3 is obviously deteriorated by the flaws in the annulus extending from ~ 30 to 60% of the total radius. Furthermore, it seems that from a qualitative point of view the generation of the rose pattern in the fracture surface is directly linked to ultimately high mechanical performance. Concluding these findings, flaws in a circle with less than 30% radius are obviously less critical for the mechanical characteristics while flaws in an annulus between 30 and 60% strongly affect the overall mechanical performance of the bonding.

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014

Downloaded/printed by

Rochester Institute Of Technology pursuant to License Agreement. No further reproductions authorized.

These results support the hypothesis that the failure of the circular bonding under tensile loading initiates within an annulus between 30 and 60% of the total specimen radius. Synchronization of the testing machine recordings with video sequences indicates that the initial rupture indeed starts inside the bonded test specimen since no cracks on the surfaces of the adhesive are visible when the load starts declining. On the contrary, when one interrupts the tensile test prior to the onset of failure, it can be shown that an inner core exists (see specimen 8 shown in Fig. 21) which is still functional. This observation led to the hypothesis of the initial break occurring in the annulus region as previously defined. One potential explanation for such behavior from a chemical point of view is the fact that diffusion of cure by-products from the adhesive to the exte­rior environment, as well as the reactive formulation components towards the curing region results in gradients in the elastomeric network density which reflect the circular symmetry of the overall test specimen. For one-component silicone adhesives such effects have been described in the literature [9] and for these materials diffusion processes are triggered by the diffusion of moisture into the adhesive bulk material; however, one may hypothesize that the occur­rence of (smaller) gradients in crosslink density even in two-component silicone adhesives are due to the diffusion of crosslink by-products from the bulk to the environment.

Furthermore, local stress loads in the bonded point specimen probably also peak upon tensile loading within the annulus previously described, irrespec­tively of the changes in crosslink density within the adhesive material. This hypothesis is supported by the fact that a similar annulus is observed in a

FIG. 21—Hypothesis of fracture surface propagation [3].

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014 Downloaded/printed by

stress-whitening, transparent silicone structural adhesive applied in bonded point supports and subjected to tensile loads at high strains [9].

In the section titled “Cyclic Tensile Loading of Point Supports" the focus was placed on cyclic testing and related mechanical performance. Figure 22 compares the mechanical characteristics of specimen 1 subjected to simple ten­sile testing (without flaws in the adhesive) with those of specimen 9, which was loaded in a cyclic manner up to 1400 N corresponding to approximately 0.7 MPa stress.

The related fracture surfaces are displayed in Fig. 23. While specimen 1 shows the regular rose pattern, specimen 9 reveals totally different surface char­acteristics. Obviously, the rather flat unstructured fracture surface of specimen 9 is linked to the poorer mechanical characteristics as plotted in Fig. 22. A potential explanation for the differences in the mechanical performance of these specimens may be the smaller surface area created by the fracture of spec­imen 9 when compared to the rose pattern observed for specimen 1. Since the creation of the free surface is linked to energy consumption, the larger the frac­ture surface, the larger the mechanical work required to generate the fracture, which is represented by the area under the load versus deflection curve. At the current stage it is unknown whether these statements hold true for other bond­ing diameters and thickness values.

In addition to the need for exploring other point support geometries in a similar manner, activities are underway for establishing a procedure dedicated to a reproducible and reliable provision of defects within the adhesive for improved quantitative control of defects in future test campaigns of point supports.

Summary and Conclusions

Tensile and shear tests were performed for varying mixing ratios and the mixing quality of the investigated representative two-component silicone adhesive

FIG. 22—Load curves for cyclic and non-cyclic tensile tests.

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014 Downloaded/printed by

FIG. 23—Fracture surfaces for cyclic and non-cyclic tests.

widely used in structural glazing. Concerning the material properties such as stiffness and strength, the adhesive shows a quite robust behavior with respect to variations in mixing ratios within the investigated mixing range. In view of the tensile and shear test campaigns, the variation of the properties with the mixing ratio is assessed to be relatively small. Furthermore, incomplete mixing, although not quantitatively judged, leads to an increased scatter of the investi­gated mechanical properties. Furthermore, the mechanical characteristics in terms of stiffness and strength are degraded due to incomplete mixing. It is assumed that the outcome of these tests can also be mapped in a first attempt on durability issues related to the mechanical properties of the two-component silicone adhesive.

In order to understand the mechanical behavior of bonded point supports under varying load schemes, cyclic test campaigns have been performed for cir­cular specimens 50 mm in diameter and 7 mm in bonding thickness. The cyclic part of the test profile consisting of 100 cycles was defined by a lower tensile force of 100 N and by upper displacements ranging from 0.25 to 1.5 mm in order to avoid backlash on the one hand, and in order to take into account the signifi­cant non-linearity of the adhesive material on the other hand. Analyses have been performed for both the cyclic part of the test curves and the post-cyclic part up to failure.

As expected, the degradation of the material depends on the amplitudes of the cycles. Two major aspects have been identified: first, the change of behavior dur­ing the cycles and second, the post-cycle characteristics with respect to the remaining load bearing capabilities. Regarding the cyclic part, the slopes decrease depending on the cycle amplitudes. For small amplitudes an asymptotic behavior is obtained, while for large amplitudes additional tests with significantly higher number of cycles are recommended in order to be able to draw final conclusions. The investigation of the load bearing capabilities after cycling showed that high amplitude cycles significantly reduce the remaining material performance.

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014

Downloaded/printed by

Rochester Institute Of Technology pursuant to License Agreement. No further reproductions authorized.

For point support specimens of the same geometry, a relationship between failure as observed in the load curve and the break in surface texture was hypothesized by analyzing the tested specimens of 5 mm bonding thickness sub­jected to tensile loading. Although special care was undertaken during the man­ufacturing of the test specimens aiming at perfect bonding application, flaws were identified in different instances, such as bubbles in the bulk material (which later showed up on the fracture surfaces). Obviously, defects located in a radial interval of 30 to 60% of the total specimen radius have a high impact on the failure behavior of the specimens while flaws located near the center of the specimen have less of an impact on the physical properties. It is unknown whether similar conclusions can be drawn for other bonding geometries, e. g., varying diameters and bonding thicknesses.

In terms of the impact on the mechanical performance of the test specimens, the impact of the investigated parameters on the durability of the bonded point supports is assessed as follows:

• Cyclic load exposure and defects of the adhesive are primary candidates in view of durability issues.

• Poor mixing quality of the investigated two-component adhesive is a secondary candidate in view of durability aspects. However, low mixing quality is difficult to quantify, so this requires the development of spe­cific metrics.

• Varying the mixing ratios of the investigated two-component adhesive within the range studied has a low impact.

Thus, it is recommended focusing future studies on the durability issues on cyclic load schemes and on representative specimens with controlled defects; first, cyclic loading schemes and controlled defects should be studied in sepa­rate campaigns and afterwards they should be applied in parallel in order to analyse potential interactions. A technical challenge that still needs to be solved in this context is the controlled seeding of the adhesive material with defects in a reproducible and reliable manner.


The writer would like to thank Dow Corning GmbH and especially Mr. Sigurd Sitte for supporting the experimental test campaigns. Furthermore, the author would like to thank the Test-Ing Material Company [3] for the successful per­formance of the test campaigns.