Abstract
Adhesively bonded joints are ubiquitous in electronic assemblies that are used in a wide range of applications, which include automotive, medical, military, space and communications. The steady drive to reduce the size of assemblies in all of these applications, while providing increased functionality, generates a need for adhesive joints of higher strength, improved thermal and electrical conductivity and better dielectric isolation. All of these attributes of adhesive joints are degraded by the presence of voids in them. The quest to minimize voids in bonded structures motivated a previous study of their formation in a solvent cast, die bond epoxy film, which undergoes a liquid phase transition during cure. That work is extended in this study by including the effects of various filler morphologies in the adhesive. Fillers are added to adhesives to facilitate handling of thin sheet formats, control bond line thickness and reduce coefficient of thermal expansion. As such, fillers are selected to be inert with respect to the adhesive chemistry, while being readily wetted by it in the liquid state. Common filler morphologies include woven and molded open meshes, fibers chopped to uniform length, and spheres of uniform or distributed diameters. Void formation is influenced by a number factors, which include wettability of the bonded surfaces, adsorbed water, amount of solvent retained in the film, volume of entrapped air, thermal profile of the cure schedule, and clamping pressure during cure. The presence of fillers in the adhesive adds the additional factors of constrained diffusion paths and increased area for void nucleation. We have changed our approach to modeling the diffusion of volatile species in adhesive joints from a finite difference calculation in a uniform adhesive medium used previously, to a finite element model of a complex diffusion space. The open source program Gmsh is used to generate the diffusion space from a set of input parameters. The calculations of concentration profiles and diffusion fluxes of volatile species at the void interface are made using the open source finite element program elmer. As done previously, the position of the void interface is updated by integrating the product of time and flux of diffusing species over the area of the interface. The internal pressure of the void is determined by application of the Young-Laplace equation, while Henry’s law is used to estimate the concentration of diffusing species adjacent to the void interface. The calculation proceeds for a time equivalent to the integral of the time temperature product required to achieve a 70% cure state of the adhesive, at which point the void interface is immobile. The experimental approach is the same as used previously, with the filled adhesive sandwiched between glass slides and cured on a hot plate while imaged through a microscope. Images are automatically captured and analyzed by using the open source program imageJ, which allows us to track the evolution of individual voids as well as the time dependent distribution of the void population. We are working to correlate these experimental results with the predictions of our finite element calculations to allow us to make insightful choices of adhesives and optimize our bonding processes.
Machine learning and finite element analysis: An integrated approach for fatigue lifetime prediction of adhesively bonded joints
