Adhesive Failure Observations

Adhesive product data sheets and the literature often cite the vulnerability of bonded joints in contact with various organic solvents. Because most adhesives tend to have polar compo- 

---

The extent of coating adhesion failure was found to be dependent upon the resistance of the polymer in the coating to hydrolysis by corrosion generated hydroxide. In this study, similar trends have been observed for adhesives. Table I shows the results of salt spray corrosion on a series of bonds between cold rolled steel adherends and adhesives of varying chemistry. The results show that there is a direct correlation between the chemistry of the adhesive polymer and the durability of the series of adhesive bonds studied. The locus of adhesion failure also appears to be related to the type of adhesive chemistry. In this study, adhesives based on polymers having a wide range of hydrolysis resistance were examined. 

With a strong interfacial bond, when a fiber fractures, the high stresses in the matrix near the broken ends are relieved by the formation of a short radial crack in the resin. There is no interfacial debonding and corresponding friction at a sheared interface, but rather, the load is transferred to the fiber by elastic deformation of the resin. The lack of adhesive failure in this case is responsible for the relatively low emission observed. 

Problems have also been observed at resist coat due to substrate nonwetting or oleophobicity. This problem is independent of the more commonly occurring pattern lifting (see Fig. 3 for examples of missing or displaced images) adhesion failure problem, which is observed after resist exposure and development. We will also focus on work addressing these latter two specific problem areas in this paper. 

Although no adhesion failures were observed for virgin aluminum substrates as found for gold with a variety of liquid silanes, VTS and vapor-phase HMDS-treated wafers were analyzed via ESCA to determine whether surface changes occurred with these treatments. For VTS-treated substrates, a dramatic 20-fold increase in the Si/Al ratio occurred with a concurrent decrease in carbon surface... 

A more dramatic failure results in peel strengths of 0-10 g/mm and is characterized as an adhesive failure at the polyimide/metal oxide interface.This was the only failure mode observed in Ti and Zr films. Isotopically tagged water used with SIMS analysis shows that on annealing water reacts with the Ti with oxygen segregating to the metal/polyimide interface and hydrogen penetrating into the bulk of the Ti, in these samples. 

Occasionally debonding was observed at the interface (Figure 8) producing smooth, dewetted surfaces on the beads. This bead-matrix adhesion failure made a relatively minor contribution to the overall fracture mode and was limited to beads larger than about 150 /xm diameter. A critical diameter range for stress distribution at the interface may exist for certain matrix compositions and volume concentrations. 

Even if the Al Oj interlayer accelerates the activation of the transverse cracking, it seems to have the opposite effect on adhesion failure. Indeed, we observe for both systems with an Al Oj interlayer (B and E) that the debonding and buckling are delayed. Therefore, the adhesion of the films is improved. The presence of this thermally grown Al Oj interlayer increases the interfacial fracture energy values to about 15 J.m in both systems. Two qualitative explanations can be proposed for the adhesion improvement. First, the Al Oj certainly permits an increase in the number of 0-Si bonds between the interlayer and the film. Second, prior to the... 

The Catastrophe. Population increases are not effected without hazard. If the initial rate is too steep or if N is too high, the sigmoid gives way to a peaked curve whose residual level is some fraction of the maximum population. This behavior, observed in latexes, has been interpreted (51) in terms of the percentage of substrate that is occupied by polymer. Most films break away from the substrate in the form of discrete Islands, reflecting the domain character of the residual stresses (in contrast with the stretched membrane analogy). Both cohesive and adhesive failures are observed. 

At this point in the analysis, it was possible to define a tolerance for bondline thickness, whereby all points within a prescribed tolerance would be considered adhesive failure, and all points outside this tolerance would be considered wood failure. Furthemtore, a second tolerance could be specified to distinguish shallow" wood failure fi-om deep" wood failure (Fig. 10). For the specimens evaluated in this study, two tolerances ( 40 pm and 60 pm) were selected for both the bondline thickness and the depth of shallow wood failure. A typical bondline thickness for block-shear specimens is about 80 pm, and the thickness of a small fiber bundle is 40-60 pm (or 2-5 fiber diameters). Table 1 summarizes the results from this analysis as well as the visual grading values obtained from the trained observers. 

Specimen 1 (Fig. 1) was an obvious adhesive failure. All observers rated this at 0 % wood failure. The laser-scan instrument, however, detected a ver>- small degree of shallow wood failure (12 %) at the 40 pm-tolerance levels. Upon closer (microscopic) extunination. it was found that there were very narrow ridges of raised radial cells running lengthwise along the specimen, which may have been detected by the laser. No deep wood failure was indicated at either tolerance level. 

Specimen 5 (Fig. 1) was selected for analysis because like specimen 1, it too is an obvious adhesive failure. However, unlike specimen 1, it contained several unique surface features that made it difficult to analyze. Among these were thick adhesive fragments, bondline voids (probably from air bubbles), and shallow (a few fibers) wood failure. Four of the observers declared near complete adhesive failure. However, one observer viewed it veiy differently (85 % wood failure). In this case, the profilometer analysis did not agree well with visual observations. This was due to considerable warp (cup and bow), which severely compromised the analysis. 

A fresh surface of thermally grown silicon dioxide, it must be pointed out, is hydrophobic. However, it quickly reacts with water vapor in the atmosphere to form silanol (Si—OH) and gradually becomes hydrophilic. In fact, the chemical vapor deposition of silicon dioxide forms only a silanolated surface. Being fairly hydrophobic, resists do not adhere well to hydrophilic surfaces such as Si02-These surfaces contain hydroxyl groups as illustrated in Reaction [11.1]. The adhesion failure of resist films on such surfaces is often observed in the course of development or wet etching. As a result, a surface treatment to promote adhesion is necessary before the resist film is deposited on such surfaces. ... 

For a wide range of materials the emission of electrons (EE), positive ions (PIE), neutral species (NE), and photons (phE) has been observed accompanying fracture. We refer collectively to these emissions as fracto-emission. In this paper we review our work on fracto-emission where the fracture event involves interfacial or adhesive failure. The interfaces to be discussed include the following brittle materials/epoxy, glass/elastomers, and brittle materials/pressure sensitive adhesives. 

This model has been tested primarily on polymer systems (a more detailed treatment will be presented in a publication in preparation) but appears to explain to a great extent the observed features of FE, in particular, the time dependence of EE and PIE. In terms of applications involving adhesive failure, such a model would be useful to relate the measured FE characteristics to fracture phenomena of interest i.e., the fragment species, density of trapped electrons, initial concentrations (before decay) and their rate of production all should be closely related to processes occurring at the crack tip. Although the FE model and characteristics outlined above have been observed in a limited number of situations, there appears to be evidence for considerable generality. 

An early discovery we made involving adhesive failure and its effect on charged particle emission concerned the fracture of composites (a more detailed discussion of this work can be found in References 13, 15, and 20). Starting with the constituents of a composite, namely, the filaments and neat resin, we found that the electron emission following their fracture appeared as shown in Fig. 4. The most important point to observe is that the emission is very rapid, decaying with time constants on the order of 10 - 100 microseconds. Early identical emission curves are observed for the PIE from the same materials. 

---