Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 STATE OF THE ART: MECHANICS The stress concentrations associated with mechanical connectors such as bolts and rivets can in part be alleviated by adhesive bonding. However, the stresses are also not uniform for adhesives, and understanding of the mechanisms of adhesive failure is far from complete. The stress and the mechanism of failure are likely to be altered and perhaps complicated by the presence of severe environments. Fortunately, recent advances in experi- mental techniques and analysis (particularly computer techniques) have provided significant information and insight into both areas. These methods should be amenable to adaptation to explore, predict, and optimize adhesive performance in hostile environments. In no area of material science is thoughtful care and planning of tests more important than it is for adhesives. Adhesive tests are used for a variety of purposes, including (1) comparing different adhesives, surface preparation methods, or curing techniques; (2) as a quality control check for a batch of adhesives, preparation techniques, effects of aging, etc.; and (3) as a means of determining parameters that can be used to predict the performance of actual practical joints. It is in this third very difficult area that adhesive fracture mechanics holds particular promise. There has been no dearth of methods proposed and developed to explore adhesives. Some of these have been formalized and published as standards by governmental agencies, the American Society for Testing and Materials (ASTM), etc. Except for a few notable exceptions, these suffer from a common problem; they can be used to predict performance only for practical joints that very nearly duplicate the test specimen's exact geometry. That is, in most standard tests, little if any attention is given to the details of the stress distribution in a joint; rather, results are almost univer- sally given as gross force per unit area. Therefore, these tests yield little if any information on extreme values of the stress or point of failure initiation. There is evidence that the reported "average stresses" may have little direct bearing on the actual failure. One of the most common and useful types of adhesive test is the single-lap shear test. Not only is it simple and economical to conduct, but 21
22 superficially it closely resembles the type of loading to which structural adhesives are often subjected in service. Care must be exercised, however, in interpreting the results and trying to infer strength of practical joints from the given test results. For example, the results of these tests are conventionally given as failure load divided by the area of overlap. Computer analysis shows that the actual stress is far from uniform and becomes much larger than this reported average near the ends of the overlap. Mathematical analyses of stress fields near the bond terminations at sharp corners imply a singularity in the stress field. The nature of the stress field is such that, even without the mathematical simplifications, very large stress concentrations exist in these regions for both elastic and inelastic materials. Perhaps more significantly, the loading of the lap shear joint induces tensile stresses that again become very large near the terminus of overlap. The relative value of the shear and normal stresses depends on details of the adhesive and adherend properties, thicknesses, etc. These tensile stresses are particularly acute for comparatively thin adherends (as generally used in ASTM tests). The evidence is very strong that failure of the joint is more closely related to these crack opening cleavage stresses than to the shear stresses (let alone the reported average shear stress). Other standard tests that might be broadly classified as either tensile or peel tests suffer from similar problems. The average stresses for loads generally reported tell little of the stress distributions and/or maximum stresses. The nature of these stresses and stress distributions is sensitive to details of adhesive and adherend material properties, geometry, sample or loading frame alignment, etc. A persuasive argument can be made that failure initiation would likely be related more to large local stresses, local stress variations, and/or flaws than it is to average values of a given stress type. Problems, such as those just outlined, with limiting stress criteria for predicting failure have led engineers and scientists to seek alternative methods that can treat singularities, using parameters obtained from a given test to predict performance of joints with different geometries. The application of the concepts and principles of fracture mechanics to adhesive systems has become quite popular. Fracture mechanics has been quite successful in solving some difficult problems in adhesive failure and has the potential of solving many other adhesive problems. The concepts of fracture mechanics should be helpful in designing tests and experiments to expedi- tiously explore environmental effects on polymers. Related procedures might help point the way toward designing joints to sustain loads while exposed to hostile environments. A few observations on adhesive fracture mechanics might be helpful. Inherent in the concepts of fracture mechanics (FM) is the concept that cracks are likely to initiate at flaws or discontinuities. In the case of bonds, additional elastically singular regions can occur at points of abrupt changes in section, geometry modulus, etc. While superficially several FM approaches appear different in concept, such approaches as the energy release rate, stress intensity factor, and J integral can be shown to be essentially equivalent (at least in the elastic case). Considerable evidence exists that adhesive failure is highly dependent on the exact mode of stress at the crack
23 (debond) tip. Modern computational methods, particularly finite element techniques, have greatly expanded the potential of FM and the geometries, joints, etc., for which it might be used. There is a wealth of FM data, methods, experimental techniques, etc., available on which to base future studies of the durability of adhesives in severe environments. Techniques have been developed to study both static failure (Gi II III) and fatigue crack growth. Tests are needed to determine the critical value of the energy release rate GC, the critical stress intensity factor Kc, or the critical value of the J integral Jc (perhaps for each of the loading modes). In principle, almost any test could be used. Certain techniques, however, enjoy certain experimental and analytical advantages. A. N. Gent and G. Hamed (1975) have made extensive and fruitful use of peel tests to analyze a wide variety of different fracture mechanical aspects of adhesives. Likewise, the blister test proposed by M. L. Williams (1969) has been adopted by W. B. Jones (1970), G. P. Anderson, B. J. Bennett, and K. L. DeVries (1979), and others to measure Gc for a variety of different adhesive systems. A tapered double cantilever beam developed by E. J. Ripling and S. Mostovoy (1971, 1974), and used by them, Bascom and co-workers (1975), and others, forms the basis of the comparatively new testing standard (ASTM D3433). Johnson and his associates at NASA have been testing and analyzing a variety of different geometries as adhesive fracture mechanics test specimens (Johnson and Mall 1984). Recently, K. N. Liechti and W. G. Knauss developed a biaxial servo-control loading device that can accurately control and independently determine the normal tangential displacement at a crack tip in a bond line with a resolution of approx- imately 10~5 in. (Liechti and Knauss 1982). Such techniques should be helpful in the study of many aspects of fracture mechanics (e.g., the effect of loading mode). It is important to recognize that many other experimental geometries could be used. Fracture mechanics should not be envisioned as a set of rules and/or sample configurations for testing and design. Rather it should be thought of more as a basic philosophy and methodology for approaching testing, design, and evaluation. It may at times be expedient to tailor-make tests for a given system or to design adhesive joints that do not closely resemble the adhesive test in which the adhesive was originally evaluated. In problems of these types, fracture mechanics has inherent potential advantages over other available approaches. Most adhesives are polymers. While linear elastic analyses might be interesting and informative, a completely satisfactory solution for the stresses, strains, energy release rates, etc., must include considerations of the time- and temperature-dependent properties of the materials. Accordingly, several investigators have made extensive efforts to include viscoelastic effects in their adhesive analyses. Particularly noteworthy are the efforts of M. L. Williams, A. N. Gent and co-workers, W. Knauss, R. A. Schapery, and H. Brinson. Much more work is needed in this area, but these studies clearly demonstrate that in some cases viscoelasticity can be incorporated in the analysis of failure and provide significant insight into the mechanical response of polymers and adhesives. Without doubt it would be advantageous to have nondestructive methods for detecting flaws such as cracks or regions of debonding along the
24 bondline. The optimal NDE test would be one in which a measurement could be quickly and easily made that does not damage the joint(s) or structure and yields results that can be directly related to the strength of the bond. A great many experimental techniques have been brought to bear on this problem, including (1) for adherend surface inspection, the contact angle, surface potential difference, and surface impedance, and (2) for NDE of the complete bond, capacitance measurement; radiographic (X-ray, y-ray, neutron, etc.) inspection; thermal inspection methods, including variations in thermal conductivity, thermal emissivity, thermal capacity, or local differences in thermal expansion; acoustic emission; acoustic inspection, such as the coin tap and its related instrumental counterparts like the Fokker Bond Tester Type I; ultrasonic techniques, including through- transmission, pulse echo techniques, resonance techniques, and spectral analysis; and holographic interference. The current state of the art for locating and determining the size of flaws is quite good and is progressing rapidly. The ability to predict the effect of these flaws on strength, debonding load, etc., is not as well developed. For a few selected adhesives and joint designs there have been some rather impressive correlations between NDE measurements and bond strength, but more study is needed before generally applicable techniques and models will be available. Most of the techniques listed here should be adaptable to the investigation of bonds intended for use in severe environments. Acoustic emission (AE) shows promise of providing a measure of the "onset" and "accumulation" of damage. G. P. Anderson of Morton Thiokol reports, for example, that in studies of debonding of a system tunnel that runs the entire length of the space shuttle, AE techniques readily detect damage at loads of 40 to 60 percent of those for which there is any visual evidence of damage (DeVries and Anderson 1979; private communication between K. L. DeVries, University of Utah and G. P. Anderson, Morton Thiokol, April 1984). In conclusion, a number of thermally and environmentally stable polymers have recently become available. It seems probable that some of these can be developed into adhesives for use in severe environments. It appears that the time is ripe to exploit some of this potential. Fortunately, experimental techniques, analytical methods, and computer techniques are available that can be adapted for characterizing these adhesives and designing for them. Experimental techniques and analysis based on the principles of fracture mechanics should be much more informative than measurements of the effect of various hostile environments on gross joint strength. With care, well-planned experiments and analysis should provide insight into the effect of these agents on more fundamental parameters, such as the adhesive or cohesive fracture energies, adhesive and adherend moduli, interfacial degradation, corrosion, or other structural changes. Models based on an understanding of the basic mechanisms responsible for behavior should provide more reliable predictions of long-term performance.
25 REFERENCES Anderson, G. P., S. J. Bennett, and K. L. DeVries. 1979. Analysis and Testing of Adhesive Bonds. New York: Academic Press. Bascom, W. B., R. L. Cottington, R. L. Jones, and P. Peyser. 1975. Fracture of epoxy and elastomer-modified epoxy polymers in bulk and as adhesives. J. of Appl. Poly. Sci. 19:2545. DeVries, K. L., and G. P. Anderson. 1979. Analysis of Design of Adhesive Bonds. Lecture Series 102, Bonded Joints and Preparation for Bonding. AGARD-LS-102. NATO. Gent, A. N., and G. Haraed. 1975. Peel mechanics. J. of Adhesion 1:91. Johnson, W. S. , and S. Mall. 1984. A Fracture Mechanics Approach for designing bonded joints. NASA Technical Memorandum 85694. Washington, D.C. : National Aeronautics and Space Administration. Jones, W. B. 1970. Cohesive and Adhesive Polymer Fracture Investigation. Ph.D. Dissertation, University of Utah. Liechti, K. N., and W. G. Knauss. 1982. Crack propagation at material interfaces: No. I experiment tech. to determine crack profiles. Experimental Mechanics 22:262. Mostovoy, S., and E. J. Ripling. 1971. J. of Appl. Poly. Sci. 15:641. Mostovoy, S., and E. J. Ripling. 1974. Fracturing Characteristics of Adhesive Joints. Final Report under contract #N00019-73-C-0163. Watertown, Massachusetts: Materials Research Laboratory, Army Materials and Mechanics Research Center. Williams, M. L. 1969. Continuum interpretation for fracture adhesion. J. of Appl. Poly. Sci. 13:29.