Statistical Characteristics of Fatigue Failure of Copper Thin Films

Abstract

Tension-tension fatigue tests were conducted on an electrodeposited copper film with a thickness of 12 mu m under four levels of maximum stress and two levels of mean stress. Statistical characteristics of the measured fatigue lives were analyzed using three estimation methods for cumulative distribution function and five probability distributions in order to identify the dominant probability distribution for the fatigue life of copper film. It was found that while the 3-parameter Weibull distribution provided the best fit for the measured data in most cases, the other distributions also provide a similar coefficient of correlation for the fit. The absence of the dominant probability distribution was discussed with considerations of the deformation mode and the scanning electron microscope (SEM) measurements of fatigue-fractured surfaces. Based on the statistical analysis, the probabilistic stress-life (PSN) curves were obtained for statistical prediction of fatigue life of the copper film in the intermediate life regime.