A State Evaluation Method for Solder Layer in MOSFET

Abstract

MOSFET is the core component in power equipment. It is widely used in electrical vehicles (EV), wind generation, rail transit and so on. The long-term impact of temperature and stress cause fatigue in the device during operation. Because of the low melting point of 96.5Sn3.5Ag, solder layer aging and failure is one of the main failure modes. So, it is important to figure out the failure mechanism and the effects of defects in the solder layer. A finite element (FE) model considered the temperature dependence of materials was built in COMSOL software to support the subsequent studies. Effects of voids in solder layer and fatigue are studied and analyzed based on the FE model. The results show the junction temperature, case temperature, on-resistance and thermal resistance between junction and case increase with the rise of voids’ areas and fatigue degree. Besides that, all of them have a similar trend, which means on-resistance can be a criterion for thorough failure replacing the thermal resistance. And the on-resistance is more sensitive than thermal resistance because its growth rate is much higher than that of thermal resistance. Based on the simulation and analyzed, on-resistance, case temperature and on-current were selected as the characteristic parameter to reflect the healthy state of MOSFET. They were used as the inputs for the evaluation model. And the growth rate of on-resistance was chosen as the output parameter. Combine the failure rate curve, the range from health to thorough failure was be divided into five pieces with different intervals. For evaluation, adaptive neuro-fuzzy inference system (ANFIS) was adopted to establish the model. By validation and comparing with some common classification algorithms, it was verified and showed high accuracy.