Predicting the lifetime of solder joints undergoing thermal cycling is crucial for the electronics industry in order to guarantee a certain performance of their products in the field. Semi-empirical methods are often used to predict the average lifetime of the critical joints. However, to get a reliable failure probability the standard deviation must also be addressed. The deviation of the lifetime from the mean value is a consequence of the variation in microstructure found in actual joints. We therefore propose a new methodology that calculates crack growth based on microstructural features of the joint. A series of random microstructures is generated. Crack growth calculations are performed for each of these structures. The structural problem is solved numerically with the extended finite element method which allows a complete automation of the process. The mean crack length and standard deviation are calculated from the crack growth simulations and the result is compared to experimental data.
展开▼