Technical Insight
Kyma reveals ammonothermal-HVPE GaN growth plans
Kyma has teamed up with Ammono and Avogy to drive novel bulk GaN growth forward. Compound Semiconductor finds out more.
Late last year, US-based wide bandgap materials supplier, Kyma Technologies, won $3.2 million from DoE agency, Advanced Research Projects Agency-Energy (ARPA-E), to develop high quality GaN substrates. Joining forces with Poland-based bulk GaN pioneer, Ammono, as well as US power electronics developer, Avogy, Kyma intends to pioneer a novel growth process to deliver the low cost, high performing GaN boules that power electronics manufacturers crave.
The project is ambitious but with Kyma chief executive, Keith Evans, describing the industry line-up as "the dream team", those involved expect success. So what lies in store?
The Kyma team will first combine Ammono's ammonothermal growth process with its HVPE-growth process to grow high quality boules quickly. Without a doubt, ammonothermal growth creates higher structural quality wafers than HVPE growth alone, but takes time. And as Kyma's chief science officer, Jacob Leach, puts it: "We want to get the structural quality of the ammonothermal process with the high growth rate of HVPE."
While Kyma's HVPE growth process is relatively fast, combining it with ammonothermal growth could boost the structural quality of wafers. [credit: Kyma]
To date, most manufacturers of GaN substrates, including Kyma, grow GaN wafers by HVPE on a non-native seed such as sapphire. But while the process is quick, it is not simply extendable to boule growth due to lattice and thermal expansion mismatch between the GaN and the non-native seed.
As Leach highlights: "There are inherent costs that you just can't get around... it just isn't a true boule approach, which is what we are aiming for here."
So with this in mind, Leach and colleagues will take a high structural quality seed GaN wafer, grown by Ammono, and rapidly grow GaN layers on it by HVPE. Their initial aim is to replicate the structural quality of the ammonothermal seed using HVPE, and then create GaN overlayers with lower impurity concentrations and point defect densities than in the original seed.
"We aim to have two inch thick boules that we've not only grown, but also demonstrated the ability to slice and polish; we want to make real substrates," says Leach. "We will deliver these to our device partner, Avogy, who will make vertical diodes using their processes, to demonstrate the utility of these substrates."
Close collaboration between the key GaN industry players offers clear appeal, but why is Kyma growing its GaN layers on an Ammono-grown wafer; why not use an HVPE-grown free-standing GaN wafer as the seed? As Leach explains, GaN seed wafers produced by HVPE on sapphire suffer from lattice lensing. Here, the crystalline lattice is curved across the wafer, which changes the local mis-cut across the wafer.
"If you measure the radius of curvature in the lattice of Ammono’s ammonothermal seeds it's of the order of around a kilometre, while the radius of curvature in our HVPE seeds is about 2m to 10m," he says. "It would be tough to take an HVPE seed and grow very thick boules by HVPE, so we want to marry these two processes."
If successful, the team intends to launch a two-inch commercially viable product by the end of the project, some four years away. Not a modest task, but Kyma chief executive, Evans, believes this is achievable.
As he highlights, Kyma is not starting from scratch, but has carried out GaN-on-GaN growth in the past. And, what's more, he is very excited about combining the ammonothermal and HVPE growth processes.
"There have been academic studies about this and Kyma has been talking about it for years," he says. "But to the best of my knowledge, this is the first time that two commercial entities have done this."
The chief executive firmly believes that high structural quality boules are crucial to driving down GaN wafer costs and provide the only route to manufacturing economies of scale. Indeed, silicon, GaAs, InP, Ge and most sapphire substrates are made by slicing from a boule.
And now he hopes the Ammono-Kyma-Avogy combo will provide the winning formula.
"We've picked the best structural quality GaN maker, plus our HVPE process and the best vertical GaN power electronics developer to make a great team," he says " You know Robert Dwilinski [Ammono] and I have been thinking about this for half a decade, and as I have said; 'they have the seed and we have the speed'."
