Energy release rates are extracted from non-linear finite element analyses of a thin film bonded to a rigid substrate that is shaft-loaded at its free edge. This geometry is of interest because it simulates a probe test that has proven to be useful in characterizing the adhesion of thin, microelectronic coatings bonded to silicon wafers. Preliminary experimental results indicate that out-of-plane rather than in-plane loading dominates failure in the system. This work therefore focuses on out-of-plane film loading. To validate finite element and energy release rate methodologies, energy release rates from finite element analyses of pressurized and shaft-loaded blister tests are first correlated to theoretical limit cases. Upon validation, mode I, mode II, and mode III energy release rates are extracted from three-dimensional continuum finite element models of the edge-loaded thin film by a three-dimensional modified crack closure method. Having assumed a circular debond as observed experimentally, energy release rates are determined by a step-wise approach around the circumference. The progression of debond is simulated in multiple analyses by altering the boundary conditions associated with increasing the debond radius. Mechanical loading is supplemented with thermal loading, introducing residual stresses in the non-linear analyses. A sensitivity analysis of energy release rates to residual stress is performed. The results indicate that inclusion of residual stress has an important role in both the magnitude and mode-mixity of energy release rates in the thin film. Increasing the length of debond effectively transitions the film from a shearing mode to a bending mode, thereby significantly impacting each mode of energy release rate differently. / Master of Science

Aromatic polyimides have found widespread applicability which can be partially attributed to their thermal stability, chemical resistance, and high glass transition temperature. However, deficiencies in their processability, solubility, transparency, and relatively high dielectric constants do not always provide the optimum properties for many specialty microelectronics applications. The incorporation of aliphatic segments to form partially aliphatic polyimides, has been used to counteract these shortcomings. Many of the potential uses of partially aliphatic polyimides require them to adhere to ceramic substrates, a main topic of this research. Polyimides and copolyimides that varied in chemical composition by their aliphatic content were characterized by their molecular weight, glass transition temperature, thermal stability, coefficient of thermal expansion, refractive index, dielectric behavior, and mechanical properties. Structure-property relationships were established. The gamma and beta sub-Tg viscoelastic relaxations were investigated to understand their molecular origins. The adhesion performance of a selected series of partially aliphatic polyimides to SiO2/Si was examined using a shaft loaded blister test, which was designed and instrumented for use in a dynamic mechanical analysis instrument. The adhesion was studied at high and low percent relative humidities and for several temperatures to examine if adhesion strength is influenced by polymer chemical composition. The adhesion energy could not be quantified for the entire series of polyimides. It was possible to interpret the quantitative adhesive fracture energies along with the qualitative adhesion strength behaviors, the failure surface analyses, and to offer an understanding of the adhesive chemical structure-physical property relationships. These understandings provide a conclusion that the incorporation of aliphatic segments into the polyimide chemical structure improves the durability of the adhesive bond to SiO2/Si under high percent relative humidities. / Ph. D.

Mechanical fatigue testing has shown that MEAs (membrane electrode assemblies) fail at lower stresses than PEMs (proton exchange membranes) at comparable times under load. The failure of MEAs at lower stresses is influenced by the presence of mud cracks in the catalyst layers acting as stress concentrators. Fatigue testing of MEAs has shown that smaller-scale cracking occurs in the membrane within these mud cracks, leading to leaking during mechanical fatigue testing and the failure of the membrane. In addition, this testing of MEAs has further established that the cyclic pressurization pattern, which affects the viscoelastic behavior of the membranes, has a significant effect on the relative lifetime of the MEA. To investigate this behavior, pressure-loaded blister tests were performed at 90 °C to determine the biaxial fatigue strength of Gore-Primea® Series 57 MEAs. In these volume-controlled tests, the leak rate was measured as a function of fatigue cycles. Failure was defined as occurring when the leak rate exceeded a specified threshold. Post-mortem characterization FESEM (field emission scanning electron microscopy) was conducted to provide visual documentation of leaking failure sites. To elucidate the viscoelastic behavior of the MEA based on these results, testing was conducted using a DMA to determine the stress relaxation behavior of the membrane. This data was then used in a FEA program (ABAQUS) to determine its effect on the mechanical behavior of the MEAs. A linear damage accumulation model used the ABAQUS results to predict lifetimes of the membrane in the MEAs. The models showed that under volume-controlled loading, the stress decays with time and the stress dropped towards the edges of the blisters. The lifetimes of the MEAs varied depending on the cycling pattern applied. This is important for understanding failure mechanisms of MEAs under fatigue loading, and will help the fuel cell industry in designing membranes that better withstand imposed hygrothermal stresses experienced during typical operating conditions. / Master of Science

Environmental fluctuations in operating fuel cells impose significant biaxial stresses in the constrained proton exchange membranes (PEM). The PEM's ability to withstand cyclic environment-induced stresses plays an important role in membrane integrity and consequently, fuel cell durability. In this thesis, pressure loaded blister tests are used to study the mechanical durability of Gore-Select® series 57 over a range of times, temperatures, and loading histories. Ramped pressure tests are used with a linear viscoelastic analog to Hencky's classical solution for a pressurized circular membrane to estimate biaxial burst strength values. Biaxial strength master curves are constructed using traditional time-temperature superposition principle techniques and the associated temperature shift factors show good agreement when compared with shifts obtained from other modes of testing on the material. Investigating a more rigorous blister stress analysis becomes nontrivial due to the substantial deflections and thinning of the membrane. To further improve the analysis, the digital image correlation (DIC) technique is used to measure full-field displacements under ramped and constant pressure loading. The measured displacements are then used to validate the constitutive model and methods of the finite element analysis (FEA). With confidence in the FEA, stress histories of constant pressure tests are used to develop linear damage accumulation and residual strength based lifetime prediction models. Robust models, validated by successfully predicting fatigue failures, suggest the ability to predict failures under any given stress history whether mechanically or environmentally induced - a critical step in the effort to predict fuel cell failures caused by membrane mechanical failure. / Master of Science

The strength and durability of proton exchange membranes for use in fuel cells has received much attention recently due to the increased push for sustainable alternatives to the internal combustion engine. To be viable, these alternatives must have comparable lifetimes and power outputs to the internal combustion engines they replace. Chemical degradation was once viewed as the most common culprit of early fuel cell failure, but as membranes and catalysts improved, mechanical failure became an important factor. As a result, fundamental research on the mechanically-induced failure mechanisms of fuel cell membranes, coupled with development and processing of less expensive membranes, has become an important topic. The use of the blister test geometry, along with digital image correlation of the deformed shape, creates a self-contained analysis tool useful for measuring the biaxial strength of membranes. In this work, blister tests are used to measure biaxial stress and strain for fuel cell membranes subjected to ramped pressure loading to form stress-strain curves that indicate the onset of yielding under biaxial stress conditions. Stress-life curves are developed experimentally for Gore-Selec? series 57 members using data collected under constant pressure conditions. These results are used to predict blister failure under ramped and fatigue loadings. A newly implemented hydrocarbon membrane system is evaluated with constant-pressure-to-leak blister testing. Improved strength following an isothermal hold at 100°C (pretreatment) is shown to occur. Ramped pressure testing indicates that the material after the pretreatment is stiffer and has a higher yield stress than the material before treatment. Morphological and constitutive characterization indicated differences in the materials that are consistent with the improved performance. / Master of Science

No description available.

Environmental stresses arising from temperature and moisture changes, and/or other aggressive fluid ingressions can degrade the mechanical properties of the adhesive, as well as the integrity of an adhesive interface with a substrate. Therefore such disruptions can significantly reduce the lifetime and durability of an adhesive joint.1-4 In this research, the durability of certain epoxy adhesive joints and coatings were characterized using a fracture mechanics approach and also by constant frequency impedance spectroscopy. The shaft-loaded blister test (SLBT) was utilized to measure the strain energy release rate (G) or adhesive fracture energy of a pressure sensitive adhesive tape. In this study, support for the value of the SLBT fracture mechanics approach was obtained. The SLBT was then used to investigate the effects of relative humidity on a model epoxy bonded to silicon oxide. Lastly, the effects of water and temperature on the adhesion of a commercial filled epoxy bonded to silicon oxide was characterized and interpreted. A novel impedance sensor for investigating adhesion was developed in a collaborative effort between Virginia Tech and Hewlett-Packard. Utilizing the technique of constant frequency impedance spectroscopy, the distribution and transport of fluids at the interface of adhesive joints was measured. A broad spectrum of adhesives was tested. In addition, the effects of hygroscopic cycling on the durability of adhesive coatings were measured for the commercial filled epoxy using the device. Lastly, recommended modifications of the experimental set-up with the new sensor are proposed to improve the technique. / Ph. D.

Water-assisted crack growth at an epoxy/glass interface was measured as a function of applied strain energy release rate, G, and temperature using a wedge test geometry. The specimens consist of two glass plates bonded with a thin layer of proprietary epoxy adhesive. The crack fronts along the epoxy/glass interfaces were measured using an optical stereomicroscope. The relationship between G and the debonding rate, v, can be measured using this method, and the threshold value of strain energy release rate, Gth, can be determined from the measured data. Two types of testing procedures were conducted in this study: ex situ, i.e., pre-conditioned wedge tests and in situ ones, in which wedges were applied before the specimens were submerged into water. A preliminary model was developed based on the thermal activation barrier concept, and allows the prediction of Gth for the temperatures beyond the testing region. Changes in interfacial strain energy release rate caused by thermal residual stresses in a triple-layered specimen were analyzed in Chapter Three. The method is based on linear elastic fracture mechanics and simple beam theory. The curvature of a bimaterial strip was chosen to characterize the residual stress in the specimen, and the strain energy release rate, caused by both tensile and compressive residual stresses in the adhesive, was derived for an asymmetric double cantilever beam (ADCB) geometry. The contribution of the thermal residual and mechanical stress to the global energy release rate was analyzed. The thermally induced energy release rate, GT, is found to be independent of crack length, but is a function of residual stress level and geometric and material parameters of the specimen. The adhesion of films and coatings to rigid substrates is often measured using blister geometries, which are loaded either by an applied pressure or a central shaft. The measurement will be affected if there are residual stresses that make a contribution to the energy release rate. This effect is investigated using analytical solutions based on the principle of virtual displacements. A geometrically nonlinear finite element analysis is conducted for comparison. Furthermore, the relationships among strain energy release rate, load, deflection, and fracture radius are discussed in detail in Chapter Four. Both analytical solutions and numerical results reveal that uniform tensile residual stresses reduce a specimen's deflection if it experiences plate behavior under small loads. However, this effect diminishes when membrane behavior is dominant. The mechanics of a single-lap joint with different boundary conditions subjected to tensile loading are investigated. Closed-form solutions are obtained for a specimen configuration considering different clamping methods. Based on the approach pioneered by Goland and Reissner, the solutions reported in this paper provide a simple but useful way to understand the effects of boundary conditions on this test geometry. The solutions in this study suggest that different grip configurations mainly affect the response of the specimens if the grip position is close to the joint edge or the loads are small. Generally, the influence caused by different gripping methods is only limited to the boundary region, and the behavior of the joint part subjected to tensile loading is almost the same as that for a simply-supported case. / Ph. D.