Thermomechanical fatigue damage modeling and material parameter calibration for thin film metallizations

Numerical fatigue damage models can help to save cost and time when studying fatigue damage in the copper metallization layers of power semiconductor devices. However, their predictive capabilities strongly depend on the parameters associated with these models. This paper presents a strategy for calibrating parameters of a numerical fatigue damage model using experimental results from thermomechanical fatigue experiments. Fatigue damage is predicted by the Fatemi–Socie critical plane method in combination with a Coffin–Manson law. Experimentally, test devices are utilized which can reproduce loading conditions that approximate those occurring in real semiconductor devices. Damage is measured in form of visible surface cracks by the means of surface imagery. Experimental results from devices with pronounced lateral thermal gradients are used for calibrating the fatigue parameters of the Coffin–Manson law. Eventually, the model with the calibrated fatigue parameters is used to predict fatigue damage for a second experiment with different loading conditions. For all investigated test devices the numerical predictions are in good agreement with experimental results. The simulations show that significantly more damage occurs in regions with higher temperatures and that the surface topology has a strong influence on local fatigue damage.