International Reliability Physics Symposium: Superior testing of SiC
Speakers at the International Reliability Physics Symposium detail the
most insightful approaches for evaluating the long-term capability of
SiC power devices.
BY RICHARD STEVENSON, EDITOR, CS MAGAZINE.
At the very earliest stages of development of any power device, the focus is always on instantaneous performance characteristics, such as on-resistance, blocking voltage and peak current delivery. But it’s not that long until evaluation expands to include insights into reliability. After all, if the device doesn’t last long enough to serve in any practical application, it will never be commercially viable.
As the device matures and volume production begins, new questions need answering when assessing reliability. For example, there is a need to know what fraction of devices will fail at any point in time. It’s also beneficial, from the perspective of the customer base, to vary the conditions of assessment to reflect different applications; and it may be crucial to pinpoint any drifts in key characteristics, even if this doesn’t lead to device failure, as it could still be a compromise too far.
All these sorts of questions are now being asked of SiC transistors and diodes, power devices that are well beyond the stage of simply being promising – they are netting billions of dollars per year, with sales ramping fast.
Insights into better approaches for assessing the reliability of SiC power devices, alongside commentary considering what traditional methods may overlook, formed the heart of several presentations at the recent International Physics Reliability Symposium.
At this meeting on 28 to 30 March, held on-line and also in person in Monterey, CA, illuminating talks on SiC included those from: ABB’s Elena Mengotti, who revealed the various tests that this company is using to assess the reliability of SiC MOSFETs from different suppliers; Kin Cheung from NIST, who explained why exceptional wafer cleaning fails to guarantee a great gate oxide; and Peter Moens from onsemi, who made a strong case for distinguishing between the lifetime of the SiC MOSFET and its useful time.