Failure precursor identification and degradation modeling for insulated gate bipolar transistors subjected to electrical stress

In driving equipment of smart factories, unexpected failures of insulated gate bipolar transistors (IGBTs) are often observed. Electrical stresses are one of the dominant causes for the IGBT failures in the field. However, there is little study about the effect of electrical stresses on the degradation of IGBTs. In this paper, we attempt to identify a key failure precursor for IGBTs subjected to electrical stresses and to model the evolution of the failure precursor. To achieve the goals, first, the main causes of IGBT failures are identified based on maintenance history, filed failure data, and experts' opinions. Second, an artificial fault injection method, i.e., electrostatic discharge (ESD), is employed to produce partially degraded (but not failed) IGBTs. The proper levels of the intensity of electrical loads (i.e., magnitude and number of the ESDs) are also determined. Finally, artificial ESD faults are seeded to IGBTs and potential candidates of failure precursors are measured. The steps are repeated until the failure of the IGBTs is observed. A relevant failure precursor is determined based on the results. A degradation model for the precursor is then built. It is expected that the key failure precursor determined in this study and the proposed degradation model can help avoid unexpected failure of IGBTs in driving equipment of smart factories.