The superior electro-thermal properties of silicon carbide (SiC) power devices permit higher temperature of operation and enable higher power density compared with silicon devices. Nevertheless, the reliability of SiC power modules has been identified as a major area of uncertainty in applications which require high reliability. Traditional power module packaging methods developed for silicon chips have been adopted for SiC and the different thermomechanical properties cause different fatigue stresses on the solder layer of the chip. In this paper, a 2-D finite element model has been developed to evaluate the stress performance and lifetime of the solder layer for Si devices, which has been validated using accelerated power cycling tests on Si IGBTs. The proposed model was extrapolated for SiC devices of the same voltage and current rating using the same solder material and the results show that under the same cyclic power loss profile the induced stress and strain energy in the die attach layer is much higher and concentrates on the die/solder interfacial area for SiC chips. Using the validated stress-based model, the lifetime can be quantified when SiC chips are used. This ability to extrapolate the available power cycling and lifetime data of silicon chips to SiC chips would be a key element for developing reliable packaging methods for SiC devices.

For silicon carbide (SiC) power MOSFETs, threshold voltage drift is a remaining obstacle in their way to the market. This study experimentally investigates the drift under dynamic or switching gate stresses. It is shown that, beside static stress, the switching events can themselves be a driving force of the threshold voltage drift. However, this happens only when the dynamic gate stress is bipolar. The study extends to show that the dynamic stress induced drift can be sustained. The findings can be used in further work for managing and coping with the threshold voltage drift in device applications.

With increasing applications of silicon carbide power MOSFETs, more attention is being paid to reliability issues, among which the long-term stability of the gate threshold voltage is of paramount importance. In this article, laboratory experiments are conducted to investigate the threshold voltage instability under ac gate stresses with different duty ratios, and on and off -state gate voltages. It is found that no prominent drift would occur even for bipolar gate stresses, as long as the off -state gate voltage is within (higher than) the critical negative bias voltage which is related to the device fabrication process. Furthermore, not only the gate voltage swing but also the gate voltage polarity will affect the speed of threshold voltage drift when it occurs. The findings are intended for better understanding and management of potential threshold voltage drift in device applications.

Silicon carbide (SiC) metal–oxide–semiconductor field-effect transistors (MOSFETs) are regarded as the key device for the next generation of power electronics. However, wide applications are hindered by the threshold voltage instability. How the threshold voltage drifts under both static and dynamic gate stress has been reported. But the underpinning mechanism remains to be revealed, which is the basis of the exploration of the application solutions. This letter is to investigate why the threshold voltage drifts. It is found that the local electric field plays the key role behind the threshold instability, which is a function of dV _GS / dt . Based on that, a physical model is proposed and experimentally verified. These findings provide not only a way to understand the mechanism but also a hint of how to mitigate the threshold instability by active gating in power electronics applications.

Discrete power devices are used in a wide range of applications some of which would benefit from device condition monitoring (CM). This paper presents a data-driven method to online evaluate the health status of the discrete devices. It is based on the detection of mold temperature by only one thermocouple, offering low cost and non-intrusive features. The approach looks into five electrical stressors and uses existing sensors to track the load variation, enabling adaptation to both linear or nonlinear load conditions of the inverter system; no additional sampling of electrical signal is required. Under the tracked electrical stresses, a back-propagation neural network (BPNN) is employed to establish the correlation between mold temperature and aging status. The overall design and implementation process, including the hardware design, model training and CM integration, are demonstrated on a 7.5 kW T-type Neutral-point-clamped (TNPC) Uninterruptible Power Supply (UPS) inverter.

As grid codes evolve, inverter-interfaced renewable generation may be required to take greater responsibility for grid support. It may be obliged to source a large current during a short-circuit fault in the grid and provide large dynamic reactive power, beyond the inverter rating, for voltage recovery after the fault clearance. In this sense, the inverter-interfaced generating system would behave more similarly to a conventional synchronous machine. This is not yet mandatory but current technology cannot readily support it if required in the future, because the power semiconductors would overheat unless they are drastically de-rated. This study proposes a method to increase the power semiconductor's transient thermal capacity, and hence its short-term over-current capability, by integrating a phase-change material inside the power semiconductor module. Experiment and thermodynamic simulation across different materials are used to illustrate the feasibility of the concept and the factors that should be considered in such a design. The study shows that 1.5 p.u. (per unit) over-current could be achieved for 30 s, which is typical of synchronous generators; and 3.0 p.u. fault current could be achieved for 3 s. A custom-made 1200 V, 50 A IGBT half-bridge module integrating a liquid metal as the phase change material is compared with a conventional commercially-available module. A validated simulation model is then used to further evaluate the proposal, regarding a 1700 V, 1000 A IGBT module for use in a wind turbine.