Research On Life Prediction Method Of High Power SiC MOSFET

With the rapid development of semiconductor technology,power semiconductor devices have become the core components of modern power electronic systems.And they play significant role for the transmission and conversion of electric energy in many applications.However,the power device is also the most vulnerable part to failure in the system.So the research of its reliability is great significance.If its useful life can be predicted in advance.And measures are take before the failure of the device.The occurrence of bigger accidents can be avoided.At present,the performance of silicon devices has approached the physical limit of their materials.And it is hard to meet the demands in the modern high-power applications.Therefore,the third generation wide band gap semiconductor materials have appeared,like silicon carbide.SiC MOSFET is a relatively mature silicon carbide power device.However,there are still few researches on its reliability,especially for the research of life prediction.Its useful life is shorter than that of silicon device and the device is more fragile.So it is more urgent to study its reliability.In view of the above problems,the main research contents of thesis include:(1)Thesis studies the failure mechanism and failure characteristic parameters of high-power SiC MOSFET.Based on the research of silicon devices and the basic structure and material properties of SiC MOSFET devices,the main failure modes and failure causes are analyzed.And the relationship between characteristic parameters and failure mechanism is explored.The feasibility of drain-source on-state voltage of SiC MOSFET devices as the fault precursor is theoretically analyzed.(2)Aiming at the study of failure characteristic parameters of high-power SiC MOSFET devices,the online extraction method of parameters for SiC MOSFET device is proposed and the design of related software and hardware are carried out.The variation situation of characteristic parameters with aging to failure of the device are obtained by the experimental methods.And the change rule of parameters is analyzed and summarized.It is verified that the on-state voltage can be used as the most appropriate parameter of life prediction for SiC MOSFET.(3)For the life prediction algorithm research of high power SiC MOSFET device,thesis proposes an improved algorithm by introducing the resampling method and Extended Kalman Filter algorithm based on the classical Particle Filter algorithm.The improved algorithm can be used to reduce the effects of particle degradation in the traditional Particle Filter for life prediction.And it can increase the accuracy of life prediction which is verified by experiments.In summary,aiming at the shortage of research in the field of life prediction for high power SiC MOSFET,thesis studies the aging characteristic parameters of SiC MOSFET and proposes the online extraction method of device parameters.For the problems in Particle Filter,an improved algorithm is used to improve the accuracy of life prediction.