Diffusion bonding is a solid-state joining process, used in a wide range of fields, allowing the joining of both metallic and ceramic materials to create complicated parts and geometries [1]. Diffusion bonded material has been found to contain inherent defects caused by the bond process that have been loosely associated to the lower tensile and fatigue strengths of the parent material [2]. The process is relatively new, and little research has been done to characterize these defects or their relationship to bond quality, with the exception of a recent study that did relate tensile strength to the bond quality [3]. Previous work resulted in successful diffusion bonding of Inconel 600, a high strength super alloy [4], optimization of bond process parameters, and the development of a method for qualitative analysis of bond quality [3]. The first goal of the present research is to develop a 2-dimensional characterization of the diffusion bonding defects for use in predicting fatigue life. Then, by treating the defects as pre-existing cracks, fracture mechanics can be used to predict life to failure based on initial material quality. A Monte Carlo Simulation will be used to capture the variability in the defects and quality of the bond. The overall goal of this research will be to enhance the previous qualitative bond quality assessment by using defect measurements taken with the SEM. This will allow for not only a more quantitative relationship between bond quality and tensile strength, but can also provide a relationship bond quality and fatigue strength. The result would lead to less of a need for destructive testing of samples, saving overall costs. The diffusion bonded defects were successfully modeled and characterized statistically by defining their, sizes, shapes, locations, and populations. Defect areas were characterized as following a log normal distribution. Although very random in shape, the defects were simplified down to elliptical geometries defined by their major and minor axis. Locations were determined to be truly random on the bonding surface. Many defects were present on the bonding surface and the population density followed a normal distribution with a high amount of deviation. The model was developed in three separate phases, each increased in complexity and accuracy of the bond surface. Each stage improved overall predictions but was still far of from the low cycle fatigue results that were being used as a comparison. It was determined that assumptions of independent multiple crack growth on the bond surface and the validity of linear elastic fracture mechanics were incorrect, causing an over prediction of fatigue life. Defect interaction and possibly analysis into elastic-plastic fracture mechanics must be undertaken to improve the predictions of the diffusion bonding fatigue life.

Library of Congress Subject Headings

Alloys--Fatigue--Mathematical models; Alloys--Defects--Mathematical models; Metals--Fatigue--Mathematical models; Metals--Defects--Mathematical models; Metal bonding; Diffusion bonding (Metals)

Publication Date


Document Type


Department, Program, or Center

Mechanical Engineering (KGCOE)


Gupta, Surendra

Advisor/Committee Member

Long, Michael

Advisor/Committee Member

Wayne Walter


Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TA483 .N69 2008


RIT – Main Campus