A Dazzling Future For Diamond GaN?
In March this year, Martin Kuball scooped a Royal Society award for his GaN-on diamond electronics breakthroughs. Compound Semiconductor talks to the UK researcher about his research, its applications, and more.
GaN-on-diamond: taking high power transistor applications to ever-higher power densities.
GaN-on-diamond electronics pioneer, Martin Kuball, hadn't intended to work with diamonds. On joining the University of Bristol, UK, nearly two decades ago, he had already spent several years studying GaN materials, including GaN-on-SiC devices, but diamond-related research was not on his agenda.
"The person that hired me was working in diamonds, and I said at the time, 'I will never do diamonds'," he says. "But now I am, which I think is probably the irony of life."
Today, the Professor of Physics heads up the Micro and Nanomaterials Group as well as the Centre for Device Thermography and Reliability at Bristol. Research focuses on GaN microwave and power electronic devices grown on SiC, silicon and synthetic diamond substrates.
Kuball and his twenty-strong team of researchers explore and develop the materials properties of the semiconductors, and apply novel thermal and electrical analysis techniques to build better devices.
And while, in Kuball's words, 'for me diamond research was accidental', the material is becoming increasingly important to the high power, high frequency GaN-based transistors that microwave designers crave.
Thanks to its high electron mobility and power density, GaN is now the technology-of-choice for RF electronics, from radar and satellite to communications and electronic warfare, and for power conversion systems.
However, progress for even higher power devices is stalling as engineers struggle to design a device to adequately dissipate heat through its thermally resistant-silicon or SiC substrate.
So instead, researchers worldwide have turned to GaN-on-diamond. As Kuball puts it: "If you push GaN-on-SiC devices too hard, they heat up too much and fail early. So researchers decided to use the 'mundane' property of diamond - its high thermal conductivity - and incorporate this to these devices."
Come the early 2000s, US-based Group4Labs, recently acquired by Element Six, had bonded diamond to GaN, with pleasing results, and so process development started apace.
In today's processes, pioneered by Element Six and other organisations, a silicon substrate is removed from a GaN-on-silicon wafer, and replaced with a 100 micron thick layer of chemical vapour deposited diamond.
Typically, with the host silicon layer removed from the AlGaN/GaN epitaxy, a thin dielectric material is first deposited onto the exposed surface, with CVD diamond then grown onto the dielectric surface.
And the results have been good. In 2013, TriQuint claimed a GaN-on-diamond breakthrough, demonstrating transistors with power densities some three times greater than the GaN-on-SiC equivalent.
Meanwhile, Raytheon has developed transistors, also based on Element Six's GaN-on-diamond wafers, that show a three times improvement in RF areal power density and nearly three times reduction in thermal resistance compared to GaN-on-SiC devices.
Professor Martin Kuball: "Working with companies is important...you really get to understand what the challenges are."
So, with the technology on the cusp of commercial success, research continues at speed. As Kuball simply states: "From a physics point of view, there's a lot of thermal transport in the dielectric and diamond layers."
"And of course, there's still a lot of engineering too," he adds. "For example, how do you modify diamond growth so the device works properly? I call this minimising the effects of the not-so-fortunate parts of the diamond microstructure and dielectric."
Right now, Kuball and colleagues are working with myriad companies and organisations, including, for example, Element Six and the US-based Naval Research Laboratory on how to seed, and grow, the diamond layers for best device performance.
His team is also exploring ways to further minimise thermal resistance between GaN and diamond layers, by, for example, altering the thickness of the dielectric layer.
GaN-on-diamond research aside, as early as 2005, Kuball was already developing techniques to probe channel temperature of GaN transistors.
GaN-based HFETs for next-generation millimetre-wave communications lacked the reliability for real-life applications, and self-heating effects - temperature increases due to Joule heating - were a key issue. The then available thermography and spectroscopy methods either lacked the necessary spatial resolution to measure peak device temperatures or were simply too slow to track thermal changes across large surface areas.
With this in mind, and working with US-based Quantum Focus and Renishaw, UK, Kuball integrated an infrared camera into a Raman microscope to investigate micrometre-scale local heating in AlGaN/GaN HFETs, with astounding results.
Using the combined Raman spectroscopy-IR thermography system, the researchers produced infrared temperature maps of devices, alongside Raman temperature linescans, pinpointing hotspots, dislocations and other imperfections, to a resolution of better than 0.5 µm.
According to Kuball, while the instrument has proven crucial to research successes, he and colleagues are still building on the original system for future GaN-on-diamond device research.
For example, reflectance measurement components have recently been added to the system, to further assess the thermal resistance of the interface between GaN and diamond layers on a wafer.
"This is for wafer-screening," says Kuball. "But we are also trying to improve the spatial resolution beyond its central half-micron spot."
Eventually, the researcher hopes to push instrument resolution to 200 nm, which would remove the need for thermal simulations for channel temperature determination during reliability testing.
Critically, this would also allow researchers to directly study inhomogeneities within GaN layers, and design more reliable devices.
Only weeks ago, Kuball won the Royal Society Wolfson Research Merit Award for his development of GaN Diamond electronics and novel thermal management concepts.
Research continues, and for him, better understanding and quantifying heat transport from GaN to diamond layers will be essential to the future of reliable devices.
As he says: "If you have this knowledge you can redesign diamond growth and the whole design of a structure to make the material even more workable."
But it's not all GaN-on-diamond. GaN-on-silicon and GaN-on-SiC research is also important, with for example, his team exploring the impact of carbon-doping in GaN-on-silicon high voltage devices.
This includes gaining a detailed electrical understanding of how the devices work, including the role of charge trapping and transport within the buffer layer of the devices, as well as the physics of device failure.
Across the substrates, Kuball's team is working with a host of companies, from NXP and OnSemi to Qorvo and United Monolithic Semiconductors. And looking to the future, Kuball expects each technology platform to be of importance in its respective application.
For example, he anticipates seeing GaN-on-silicon, in low-cost power electronics application for some time yet.
And as he adds: "GaN on silicon carbide works very well, we've had a hell of a lot of investment here, while for GaN-on-diamond we already have reasonably priced, good-sized wafers."
"These things always happen in parallel but working with companies is important," he says. "You get access to really good devices and you really get to understand what the challenges are."