Commercialising Diamond GaN
Can a multi-million pound UK project unlock Diamond-on-GaN's full potential, asks Compound Semiconductor.
GaN-on diamond HEMTs can reliably reach higher power densities than today's GaN-on-SiC devices.
As the semiconductor industry's need for speed and energy efficiency continues unabated, global demand for the mighty GaN transistor is rising.
Thanks to high electron mobility and power density, GaN is the material of choice for RF electronics applications such as radar, satellite and electronic warfare.
But power comes at a price, and in this case, it's heat. While GaN-based HEMTs have reached stupendous RF power levels of up to 40 W/mm at frequencies exceeding 300 GHz, self-heating means devices can only be reliably operated to power densities of 10 W/mm.
However, a UK-based multi-million pound research project now looks set to realise the full power potential of GaN swapping the substrate from SiC to diamond and integrating novel cooling to boost power performance by up to a factor of six.
Launched early this year, 'Integrated GaN-Diamond Microwave Electronics: From Materials, Transistors to MMICs' will develop GaN-on-Diamond HEMTs and MMICs in a bid to banish the thermal management issues that limit current GaN electronics.
Diamond has an incredibly high thermal conductivity, and thanks to this property, a diamond substrate has the potential to boost heat spreading in a transistor by up to six times compared to the SiC equivalent.
Project leader and physicist, Professor Martin Kuball from the University of Bristol, has been developing GaN-on-diamond devices for several years, and is confident the time is now right to take the technology further.
"Many industrial partners want higher power microwave electronics devices but are limited mainly by heat extraction from the electronics." he says. "We've been involved in many programs to develop GaN on diamond for this reason, and I now believe there is a huge opportunity for the UK to pull off a full integrated program here."
Indeed, the project, funded by UK-based Engineering and Physical Sciences Research Council (EPSRC), has attracted myriad industry players including MACOM, Plessey Semiconductors, IQE, Element Six, Logitech and the National Microelectronics Institute.
The Universities of Bristol, Cardiff, Cambridge, Birmingham and Glasgow are all on board. And Kuball also highlights how aerospace interest is rising, bringing the European Space Agency, as well as Airbus Defence and Space aboard.
Crucially, many Europe-based industry players are concerned that the US International Traffic in Arms Regulation (ITAR), which controls the trade of defence-related products and services on the United States Munitions List, will restrict market access to state-of-the-art GaN microwave technology.
"I envision that a GaN-on-diamond MMIC would fall into ITER's strike, so getting this technology into Europe would involve a lot of paper work and be very painful," says Kuball. "There are many challenges to tackle but by having control over the full supply chain, from epitaxy growth to application, we can more easily bring in new device design ideas."
A key part of new device design is to incorporate novel cooling systems to the GaN-on-diamond chip. As part of a US Defense Advanced Research Agency program - Intrachip/Interchip Enhanced Cooling (ICECool) - industry heavyweights Raytheon, Northrop Grumman and BAE Systems have already been devising different microelectronics cooling set-ups for such devices.
Raytheon, for one, has developed a systems of coolant-filled high-aspect ratio diamond microchannels that act as a heat exchange. These microchannels channels are positioned next to the known hot spots within a GaN-on-diamond MMIC to boost thermal performance.
Meanwhile, Northrop Grumman lines microchannels etched into the base of a GaN-on-SiC chip with a diamond coating. These channels whisk away heat and reduce heat flux by spreading it across a relatively large area.
According to Kuball, the latest UK project will try out new approaches. The researcher will not be drawn on the detail but highlights how the project team will optimise diamond thermal conductivity close to the active GaN device area.
For example, in today's GaN-on-diamond devices, a thin dielectric layer is deposited onto the GaN surface to enable seeding and deposition of diamond onto the GaN. Unfortunately this GaN-dielectric-diamond interface has a poor thermal conductivity.
So, given this, Kuball and colleagues first hope to reduce thermal resistance by eliminating this seeding layer as well as optimising the nearby diamond grain structure. "Our team has some very good diamond growers, including researchers at the University of Cardiff and Bristol, and we believe we can avoid using this seeding layer," says Kuball. "This needs to be tested but we strongly believe we can do this."
In a similar vein to past US research, the team also intends to investigate the use of microfluidics, but this time, will introduce novel phase change materials into the channels to boost heat removal further. Kuball describes this as a 'dramatically more powerful approach than conventional microfluidics'.
"If you use a phase change material, additional energy is required for the phase change to take place, improving the heat-carrying capability of the liquid in the microchannel," he says.
According to Kuball, his team has a range of methods to introduce the microfluidic channels to the epitaxy layers, and adds: "We have a few other ideas on how to construct the walls of the microfluidic channels which are different to what has already been done."
So what now? Within the initial two years of this five year project, Kuball hopes to have optimised epitaxy growth, developed the novel microfluidics, experimented with new design concepts and fabricated working devices. "We hope to have delivered a few early trial devices within year two," he adds.
And come the end of the project in 2021, he expects a fully optimised device to be in place and also hopes to have formed a spin-out company with industrial partners.
"We have a multitude of challenges to face from the seeding layer to the integration of the microfluidics. Also managing stresses and wafer bow is not going to be trivial," he says.
"But as I think is quite evident, I am very excited to be running this programme," he adds. "We have such a good team of industrial supporters and this is so critical when developing a new technology."