Electrical Trees Growth and Self-healing Properties in Silicone Gel Encapsulation Materials for Power Module at High Frequencies
Silicone gel is widely used in the package insulation of power module because of its good electrical,thermal and mechanical properties.Electrical trees are dendritic discharge channels due to partial discharges,constituting a prevalent failure mode in power module package insulation systems.On this basis,this paper builds an experimental platform for the growth of electric trees in silicone gel under high frequency sinusoidal voltage excitation,conducts experiments on the growth of electric dendrites at different frequencies,investigates the initiation and propagation mechanism of electric dendrites in silicone gel under high frequency,and analyzes the influence mechanism of frequency on the characteristics of electric trees.It also conducts experiments on the self-healing of electric trees at different temperatures and group distribution ratios,and investigates the self-healing behavior of electric trees in silicone gel and its influence mechanism.The self-healing behavior and its influence mechanism are investigated.The experimental results show that with the increase of frequency,the electric trees morphology in silica gel will be more complex and the starting voltage will also decrease,and the starting voltage at 18 kHz is about 17%lower than that at 8 kHz.The analysis suggests that Maxwell stress is the main reason for the initiation of electric trees at high frequencies,and the propagation characteristics of electric trees in silicone gels are different from those of conventional solid dielectrics.For the self-healing characteristics,the increase of temperature is favorable to the self-healing of the electric trees,but it will inhibit the self-healing at 100°C.The analysis obtains that the temperature affects the different stages of self-healing by influencing the gas state.Also,the composition ratio affects the morphology of the silica gel and the electrical tree properties.
国家科技期刊平台
NSTL
万方数据
