Diamond delivers a hike in power density

Northrop Grumman’s GaN HEMTs deliver a five-fold increase in RF power densities, thanks to the exceptional thermal conductivity of polycrystalline diamond.

BY RICHARD STEVENSON, EDITOR, CS MAGAZINE

Our industry does not rest on its laurels. Following the emergence of any ground-breaking device, it’s not long before efforts are directed at finding ways to deliver a hike in its performance.

For the RF GaN HEMT, DARPA is driving improvements in key metrics through a programme entitled THREADS: Technologies for Heat Removal in Electronics at the Device Scale. Within this programme, diamond is viewed as the ultimate material for optimising heat extraction, due to its incredibly high thermal conductivity – it’s five times that of copper.

Thanks to diamond’s exceptional thermal conductivity, devices drawing on this attribute run cooler, opening the door to operation at far higher power densities. That’s valued in many applications, including advanced radar, satellite communication, several forms of secure communication, electronic jamming and very high-performance sensing. Armed with more powerful devices, designers can construct systems that deliver an increase in performance; or maintain performance while using fewer devices, leading to a trimming size and cost.

A key player in the THREADS programme is Northrop Grumman, which is working with engineers at Stanford University.

Efforts at Northrop Grumman began in December 2023, supported by $6.4 million from DARPA, which financed the fulfilling of the key objectives of the first phase of the project. This includes propelling the power density for GaN-on-SiC HEMTs operating below 10 GHz from a previous benchmark of around 5-7 W mm^-1 to around five times this figure. That’s an ambitious goal to set – but DARPA is well-known for instigating programmes with very challenging objectives.

“We were able to achieve 26.3 watts per millimetre,” remarks Ben Heying, director of microelectronics at Northrop Grumman's Space Park Foundry, who explains that this success stemmed from trimming thermal resistance. The team dropped the thermal resistance by ‘0.37X’ – exceeding the ‘0.40X’ target reduction.

Heying is keen to emphasise another important success realised in phase I, which is reliability.

“When you're powering a device at these high powers, they typically don't last too long, because the temperature gets very, very high. But we were able to achieve 1,000 hours, so very good stability.”

This matters to Northrop Grumman, which has little interest in ‘hero’ results.

“We want to get something that we can actually put into our systems,” explains Heying, pointing out that this implies strict reliability requirements.

Not included in the objectives of the THREADS programme is a requirement for the maximum operating temperature of the HEMT. But that does not mean that team at Northrop Grumman doesn’t record this.

Heying says that throughout the programme he and his team have been developing thermal models that account for different materials and measure thermoreflectance – the temperature changes inferred from how a material reflects light to gauge temperature of the device.

Cooler diamond
Key to Northrop Grumman’s success is the deposition of diamond at a low temperature, a step undertaken at Stanford University. Deposition occurs after forming the HEMT and its accompanying circuitry, and thinning the SiC substrate to around 100 µm.

While the growth of diamond is carried out in a commercial reactor, the process is not trivial. Heying describes the details as ‘very proprietary’.

Using a low growth temperature avoids damage to the HEMT, but risks the deposition of graphite, which has an inferior thermal conductivity. Engineers at Stanford take care to avoid this, growing high-quality grains with a diameter of the order of 50 µm.

To optimise HEMT performance, engineers form a channel in the SiC along the width of the gate, which is the hot spot for this transistor. Into this trench goes diamond, to ensure that the device benefits from a very low thermal resistance. The transistor is designed so that following extraction of heat from the hottest region, lateral spreading takes place, to enhance thermal management.

Working in partnership with Stanford University, engineers at Northrop Grumman have developed more power RF HEMTs, which feature exceptional heat extraction close to the gate, thanks to the addition of polycrystalline diamond.

Phase II
Recently, DARPA awarded Northrop Grumman an additional $7 million to participate in phase II of the THREADS programme, running until the end of 2027. Efforts are directed at delivering an additional three-fold increase in power density.

Heying is not divulging details surrounding how the team will tackle this next goal, but does reveal that efforts will be directed at optimising the handling of the electrical fields within the HEMT, and how the materials come together.

“The pure materials that we're working with have a fairly good thermal conductivity, but we're putting multiple layers together, and every time you do that you end up with an interface. So, you have to engineer those as well.”

From R&D to production
It will not be long before Northrop Grumman’s success in the THREADS program benefits its customers.

“Over the next couple years, probably a year and a half, we're going to be taking a variant of that Phase I device and turning that into something that we can put into production.”

What’s not clear is whether the new commercial products will feature diamond-based cooling. Note, though, that Northrop Grumman has this capability in-house, and could transfer technology from Stanford.

So, many pure-play commercial customers, as well as those working in the defence sector, could soon benefit from the breakthroughs at Northrop Grumman. This US-based company has partners all around the world, and so as long as they are granted an export license, they will be beneficiaries from more powerful HEMTs with superior cooling, possibly thanks to the addition of diamond.