A Study On The Lifetime Prediction Technique For The IGBT Power Module

As a new type of power switching devices, Insulated Gate Bipolar Transistor(IGBT) is widely applied in new energy power generation, high voltage transmissionand other key fields for its many advantages, such as high current ratings and lowsaturation voltage, etc. However, due to its operating in high voltage, high current andhigh-frequency switching state for long term, fatigue and aging are always caused byfrequent fluctuation of power dissipation and junction temperature. more work has beendone to investigate the failure mechanism caused by temperature cycling, and based onwhich the lifetime prediction technique is being actively developped to enhance theoperating reliability of converters. This thesis focuses on the research of electro-thermalmodeling, junction temperature assessment and lifetime prediction of IGBT.Firstly, the basic structure and operating characteristics were introduced; then, twotypes of failure modes for IGBTs were summarized, and followed by an analysis ofphysical mechanism of the bond wire lift-off and solder fatigue in IGBT modules withthe thermal-mechanical point of view, which revealed that temperature is a major factorof IGBTs’ reliability, and laid a foundation for the junction temperature assessment andlifetime prediction using electro-thermal model.Then, a electro-thermal model, based on average power losses in switching periodwas presented combined the requirement of lifetime prediction with the existing models,the model, which could display the power losses and junction temperature waveformsvaried with load current, was validated by a comparasion of data calculated using theSemiSel supplied by device manufacturer，and showed its correctness and usefulness,and was suitable for lifetime prediction application.Then, in order to realize lifetime prediction by suing junction temperature profile,the overall idea of lifetime prediction, based on Pamgren-Miner damage theory andrain-flow counting method, was determined, and the rain-flow counting program wasachieved. Then, take a three-phase inverter used in electric vehicle for example, thestatistical characteristic of junction temperature was extracted by using rain-flowmethod and then life prediction of power module.was realized by combining withfatigue damage theory.Finally, by the simulation, the influences of ambient temperature and switchingfrequency of power module on lifetime and the influence of internal state (welding layer aging) on lifetime prediction accuracy were analyzed. The results show that: theincrease of ambient temperature results mainly in the increase of the average junctiontemperature, but reverse for lifetime; The increase of switching frequency results in theincrease of mean and range value of junction temperature, Resulting in short lifetime;while the junction temperature estimation error is caused by ignoring the fatigue ofwelding layers, which leads to an obvious error in lifetime prediction value. So, newlifetime prediction models should be studied to improve lifetime prediction accuracy.This study dealt with the electro-thermal modeling, fatigue damage and lifetimeprediction of IGBT modules, which would be valuable for the research onelectro-thermal design, reliability assessment and on-line lifetime prediction of IGBTs.