Research On Power MOSFET Model For Reliability Analysis

With the characteristics, such as simple drive-circuit, fast switching speed andease of integration, power MOSFETs are widely used in the power supply, electricautomobile and aerospace systems. As the continuous improvement of process andthe reduced development cycles, the reliability of power MOSFET is undertremendous test. The traditional reliability research of discrete device wasestablished on accelerated life test (ALT), however, ALT did not implement the―true accelerated‖. Therefore, there are important theoretical significance andpractical value to study the power MOSFET’s reliability, which was based on finiteelement modeling.On the basis of the thorough study of power MOSFET’s failure modes, theoverall research program of reliability modeling was proposed. First, the finiteelement model for the Power MOSFET with TO-247package was established inANSYS software, and the temperature distribution under the electric field can beobtained. Compared to the actual value shooting by the thermal imager, thesimulation result of established model was proved. Then, the device was put into theactual operating conditions to acquire the most prone parts to failure. Studies hadshown that the parts of power MOSFET with the lowest reliability were the adhesivelayer, chip, bonding wire and plastic packaging.According to the environmental factors for each failure mode, the analysis ofthe main failure modes was carried out respectively. Combined with the Parisequation, Coffin-Manson’s law and other theoretical models, the remaining life ofthe device can be predicted by the temperature and equivalent thermal stress results.Finally, reliability tests of20power MOSFETs under temperature cycling,power cycling, thermal breakdown and hot/humid environment were accomplished.Critical electrical parameters were measured by power device analyzer before andafter the reliability test, and optical microscope and SEM were used to observe thefailure morphology of device without the package. The influence by each failuremode can be confirmed by the morphology, and be effectively analyzed by theelectrical testing results, which also verified the forecasts of the remaining lifeindirectly.