Fracture and residual stress measurement in polycrystalline CVD diamond

Polycrystalline CVD diamond disks, supplied by Texas Instruments Corp., Norton Co., and Raytheon Inc., were fractured in ball-on-ring loading. Fracture surface analysis was performed on the fractured specimens using both optical and scanning electron microscopy. Fracture surface analysis was used to locate and measure the size of the failure initiating critical flaws in over 60 specimens. The flaw shape (a/b ratio) distribution suggested that the cause of the strength reducing flaws was mechanical in origin. The main part of the fracture study concentrated on additional specimens sent by Texas Instruments in which twenty different "process" conditions were examined for strength and fracture toughness. The study offered an unprecedented opportunity to study fracture in CVD diamond with a large sample set. Through the examination of the fracture surfaces the fracture path in the samples was demonstrated to be mostly transgranular. The combined data from the different sources showed that different processing conditions could lead to widely varying fracture toughness values. The strength was found to vary strongly with thickness in the Texas Instruments study whereas the fracture toughness, K{dollar}sb{lcub}rm Ic{rcub},{dollar} varied in a random manner with sample thickness.; Residual stresses in diamond coatings deposited on (110) surfaces of single crystal Silicon by hot-filament CVD were measured. The residual stresses in seven samples were measured using the sin{dollar}sp2psi{dollar} X-ray diffraction method and verified using a curvature technique. The coating thicknesses were varied between samples from 2.5 to 15{dollar}mu{dollar}m. Residual stresses in the coatings were found to be in the range 65-355 MPa and were found to decrease in magnitude (less tensile) with increase in thickness over the range of thicknesses examined. This is opposite in sign and different in magnitude from that predicted by thermal expansion mismatch. The curvature of one of the samples was measured over the cooling temperature range from near the deposition temperature (850{dollar}spcirc{dollar}C) to room temperature. The resulting stress (calculated from the curvature) vs. temperature curve followed that expected from thermal expansion mismatch but was shifted by a constant, temperature independent, value of about 520 {dollar}pm{dollar} 75 MPa.