Defect tolerance assessment of silicon nitride in rolling contact.

Abstract

This thesis focuses on two relatively unexplored types of surface defects on silicon nitride balls – ‘star’ type defects and ‘missing material’. The main objective of this research is to determine failure modes and the critical or tolerable defect sizes for rolling bearing applications. This is achieved by means of both experimental and numerical techniques (for finding crack initiation location). A modified four-ball machine is used for the rolling contact fatigue experiments on star features and missing material defects (both during manufacturing process and artificially produced) on the surface of silicon nitride balls. Experiments are conducted at different loads (3.8-4.8GPa), lubrication types, specific speed (7500rpm) and temperature conditions (75°C) to find rolling contact fatigue limitations in relation to these defects. Laser micromachining is used to produce precise holes and trenches to simulate missing material. Two different types of bearing grade silicon nitride are tested to determine tolerance and failure modes in rolling contact fatigue. Scanning electron microscopy, light microscopy and white light interferometry are used for surface analysis and topography. Post-experiment analysis in the star feature experiments has shown that star like radial cracks are prone to develop into missing material by internal fracture over the extent of the star in lubricated rolling contact. Lubrication quality or film thickness and orientation of the pre-existing cracks to the rolling direction influenced the damage process and severity. In the case of missing material experiments, samples with different diameters (50-100μm), depths (5-50μm) and shape (right cylindrical, conical and oblique cylindrical) were tested. It has been confirmed that, apart from the main experimental parameters like applied pressure, cavity diameter and depth, parameters such as cavity base profile, shape and cavity location and orientation on the contact track are important for rolling contact fatigue of silicon nitride material. Replica produced cavity base profiles to investigate failure reasons. Cross-sectioning also gave very good insight of surface and subsurface features both before and after testing (including un-failed specimens). Incipient spalls on tested samples provided the information to understand the failure mechanism (mainly spalling) in the material. FEM is used for finding stress fields at surface and sub-surface positions and ultimately predicting the location and position of crack initiation. 2D (Axi-symmetric and plane strain) and 3D models are developed to compare the results, whereas static versus quasi static analysis is presented to examine the effect of rolling. Models are verified using classical contact theory. Hydrostatic pressure effects are also successfully modeled using the finite element analysis approach. Fluid elements modeling on the solid interface of the cavity produced a different stress field and have indicated the location of crack initiation.