Serving the Army with 10 kV SiC IGBTs

Over the next two years, Navitas will develop military-grade 10 kV SiC IGBTs and accompanying p-i-n diodes.
BY RICHARD STEVENSON, EDITOR, CS MAGAZINE

Within the pantheon of pioneers of SiC power devices operating at very high voltages, Navitas stands head-and-shoulders above its peers. Its triumphs include: the first 6.5 kV SiC thyristors, unveiled in 2010; the first 10-15 kV p-i-n diodes, reported in 2012; and the launching of the first commercial 6.5 kV SiC MOSFETs in 2021.

In addition to these breakthroughs, it has a track record delivering military-grade reliability.

“We were the first in the industry to come up with this AEC+ reliability standard, which goes above and beyond the AEC-Q101,” remarks Navitas’ CTO, Siddarth Sundaresan.

So, with these compelling credentials, combined with manufacturing on home soil, it’s easy to understand the US Army’s decision to select Navitas as a partner, funded to develop military-grade 10 kV transistors.

Navitas will work towards this goal in a 24-month project entitled ALATTIS – that’s short for Accelerated, Large-Area, 10 kV SiC IGBT.

The US Army is keeping its planned uses for these power transistors under wraps. However, it would be reasonable to assume that they will be deployed in situations where very high powers must be converted, delivered or transmitted by systems with a small footprint.

Navitas claims the highest volts-per-micron for a 15 kV SiC MOSFET

What’s in a name?
Included in the title for the project is the word ‘accelerated’. It is included to emphasise the need to speed the transition between producing a device in a lab, and having a process for manufacturing in volume.

“There's literature on high-voltage IGBTs, even up to 27 kV, for a decade plus. But there is really no domestic process to manufacture them, make them repeatably,” points out Sundaresan.

Researchers reporting headline-grabbing blocking voltages have enjoyed success with devices with small areas – this helps avoid issues associated with defects. But real systems need bigger devices, capable of carrying higher currents, hence the inclusion of ‘large-area’ in the project’s name.

Producing large-area devices – that’s having dimensions of at least 6 mm by 6 mm – is a challenge. As well as avoiding defect-related issues, efforts must be directed at ensuring high-quality edge termination.

For the ALATTIS project, work focuses on the IGBT, rather than MOSFET, as the latter would require a very thick, lightly doped drift layer to block 10 kV. While growth of this layer is possible, the device would suffer from a very high on-resistance, as this scales with blocking voltage by a power of 2.5.

One merit of bipolar IGBTs, compared with unipolar MOSFETs, is conductivity modulation in the drift layer. It is this strength that makes the IGBT the preferred switching device for blocking voltages in the 10-15 kV range.

According to the Navitas CTO, the downsides of the IGBT are a heightened sensitivity to basal plane dislocations and relatively slow switching. To minimise these issues, efforts are directed at realising a high epitaxial quality, and carrier lifetime engineering.

From MOSFETs to IGBTs
Despite the weaknesses of the bipolar devices at very high voltages, the roadmap for the design-fabrication-test iterations begins by targeting 10 kV MOSFETs on 150 mm wafers.

“If you look at the processing process steps for a MOSFET compared to an IGBT, they are about 90 percent the same,” says Sundaresan.

This strong overlap will lead to valuable insights that aid the development of 10 kV IGBTs, including understanding how thicker epilayers impact bow and warp issues, potentially exacerbated by high-temperature implants.

When the engineers at Navitas shift from MOSFETs to IGBTs, they will continue with their trench-assisted planar technology.

According to Sundaresan, this technology offers the best of both worlds, combining the gate oxide robustness and long-term stability associated with the planar gate architecture with the higher cell density and lower on-resistance of trench devices. The addition of the trench allows engineers to shape and shield the electric field from the gate oxide interface.

Due to demand for defence-grade reliability – failure is not an option – Navitas’ engineers must produce devices that can withstand decades of operation under a wider range of temperatures and extreme thermal cycling. In addition, evaluations of the IGBT will consider cosmic ray-induced failure, partial discharge, and humidity.

“All of these kinds of things are a lot more exacting at 10 kV, compared to say 1200 volt or 1700 volt,” says Sundaresan, who explains that what’s critical is not just being able to withstand one extreme, but a combination of factors that stress a device.

To accompany the 10 kV IGBTs, p-i-n diodes will be developed through the ALATTIS programme. Efforts will draw on Navitas’ previous work with this class of diode. “The difference is we now need to match the p-i-n diode performance to the IGBT,” says Sundaresan.

For the development of 10 kV IGBTs and matching p-i-n diodes, Navitas will have to work with production partners, a modus operandi that some may view as hampering progress.

Sundaresan disagrees. He argues that one should not view Navitas as fabless, preferring descriptions such as an integrated fabless manufacturer; he points out that the company owns key technology stack-like device designs, key process modules, and the trench-assisted planar technology process and architecture; and he highlights the deep collaboration with foundry partners.

Sundaresan stresses that Navitas has formed strong strategic partnerships with key foundries under their integrated fabless manufacturer model. “This goes well beyond just placing wafer orders.”

Navitas claims industry-leading avalanche robustness for its 6.5 kV SiC MOSFETs

The impact of 10 kV IGBTs
If ALATTIS is successful, it will deliver transformational benefits, given that the incumbent IGBT technology, based on silicon, is limited to 6.5 kV.

For today’s 10 kV-class converters, circuits within them require stacked silicon IGBTs, voltage balancing and additional gate drives, resulting in what Sundaresan describes as a ‘clunky’ solution. As well as eliminating the need for stacking transistors, moving from a silicon IGBT to one made from SiC leads to much thinner drift layers, less charge stored, and the opportunity to switch faster. The latter opens the door to trimming the size of the magnetics, and reducing cooling demands.

These attributes are not just attractive to the US Army. They are also appealing to the designers of electrical grids, AI data centres, and solar and wind farms.

History has plenty of examples of the US military proving the viability of compound semiconductor device technologies that have gone on to achieve commercial success. And there’s good reason to think that this could be the case again with ALATTIS.