New GaN Centre To Drive Devices To Market
Germany-based researchers and industry players join forces to commercialise GaN-on-GaN research. Compound Semiconductor finds out more.
Thanks to a weighty €1.6 million in funds from the German Saxony State government, Dresden Technical University spin-off, NaMLab, and GaN substrate developer, Freiberg Compound Materials, have just launched a GaN research laboratory.
Drawing on this and additional funds from the German research ministry, the Gallium Nitride Centre researchers will now collaborate with colleagues at the Freiberg University of Mining and Technology to drive GaN-on-GaN optoelectronics and power electronics to commercial markets.
"We don't claim that we're going to compete on cost with GaN-on-silicon," says Professor Thomas Mikolajick, scientific director at NaMLab. "But we want to go into performance corners and special applications where you really need a low defect density."
NaMLab is known for its strong device development while keeping a close eye on commercial applications. However, Mikolajick is keen to highlight how the organisation will now be working closely with Freiberg Compound Materials on materials growth.
"We're going to make high quality GaN substrates," he asserts. "We're going to start with two inch wafers to raise quality and throughput levels to commercial standards, and then we will bring costs down by moving to larger diameters."
The partners will use HVPE to grow GaN substrates, growing a thick GaN layer on a template that will later be removed leaving the boule from which to slice GaN wafers.
"I can't be specific but we are using a template as the starting process, and we are not looking at ammonothermal growth," says Mikolajick. "The challenge will be to make the GaN as thick as possible so we can get a lot of wafers."
According to Mikolajick, his team will first develop high quality GaN substrates for laser applications, but more applications will follow.
"Maybe we'll even aim for high brightness LEDs with a high current density where you cannot get the necessary reliability from substrates with many defects," he says. "But general lighting applications are not on our roadmap as we wouldn't get to market for cost reasons.".
The team is also looking to develop vertical power devices, although Mikolajick admits this is very much research in progress right now. In past projects, NaMLab has defined a front-up process flow for planar GaN Metal–insulator–semiconductor high-electron-mobility transistors (MIS-HEMTs), but as part of the Saxony GaN Research Centre will now explore new architectures.
"Vertical GaN MOSFETs and HEMTs have been proposed in literature, and we are currently thinking of integrating a HEMT into a vertical configuration," he says. "This is also on our roadmap, but is still a vision."
Still, vision or not, the new GaN Centre is hardly starting from scratch. NaMLab was originally set up as a joint venture between TUD and Infineon Technologies memory products, later to become memory giant Qimonda, Germany.
NaMLab's first years were spent focusing on developing high-k dielectrics for capacitors in dynamic random access memories. But then Qimonda filed for bankruptcy in 2009, and NaMLab started looking for new markets to apply its device expertise.
Come 2010, the organisation had teamed up with Freiberg Compound Materials, with a view to accelerating the development of GaN substrates by providing its industrial partner with feedback on device performance.
"Originally we were silicon guys, with our biggest knowledge being how to integrate semiconductor devices into silicon," explains Mikolajick. "We'd done a lot of work on integrating high-k dielectrics into capacitors and transistors but discovered that GaN [businesses] were also interested in integrating [these materials]."
The organisation has since worked closely with various partners on GaN HEMT processes as well as novel device development, and is now keen to swiftly turn research results into commercial applications.
Mikolajick will not be drawn on when the first real application will surface, giving several years as a ball park figure but says: "At the end of the day, if you want to sell something, you have to understand what happens after processing and make sure you can fabricate good devices. We make simple, but complete devices where you can see the entire fabrication process."
And he is confident the strong focus on device integration will bring more GaN-on-GaN devices to market.
"You know in Germany, this is not a natural way for a research organisation to work; hand in hand with product development and commercialisation," he says. "But we are working very closely with Freiberg Compound Materials, and this is an important part of founding our research centre in Freiberg."