A novel crystal growth process could ease GaN substrate production bringing brighter, more efficient LEDs. Compound Semiconductor talks to the researchers behind the hope.

Soraa's MR16 flagship product, the GaN on GaN LED

As manufacturers of LEDs strive to deliver cheaper yet more efficient lamps, many in the industry doubt whether structures grown on sapphire or silicon substrates will make the grade.Complex epi-layer stacks with high defect densities, caused by crystal lattice mismatches between the substrate and GaN-based active layers, don't exactly point the way to cheap and easy manufacturing. And this doesn't even factor in the tricky issue of handling large diameter wafers bowed by thermal expansion mismatches.

So what to do? According to a niche, but growing, band of industry players, LED manufacturers should look beyond non-native substrates as well as the conventional vapour phase epitaxy processes used to grow LED layers on them.

Instead, high current density GaN structures could be cost effectively fabricated on quality GaN substrates, which, crucially, have been grown via a novel, but established process called ammonothermal crystal growth.

California-based Soraa is exploring this strategy and was recently selected by the US Advanced Research Projects Agency-Energy (ARPA-E) to lead a project to develop bulk GaN substrates. The LED manufacturer already sells LED-based lamps fabricated on the same HVPE-grown GaN substrates used in laser diodes for Blu-ray applications, but hopes to cut manufacturing costs and raise substrate quality with its ammonothermal growth technique.

The technique is based on an analogous method used to grow synthetic quartz crystals, in which seed crystals are crystallised within thick-walled steel autoclaves that withstand the high temperatures and pressures required for crystal growth.

As Soraa vice president of bulk technology, Mark D'Evelyn, says: “[This growth process] is highly scalable and inexpensive, bringing the promise of excellent manufacturability and crystal quality to ammothermal GaN crystal growth.”

However, growing GaN crystals in this way demands higher temperatures and pressures, stretching your standard steel pressure vessel beyond its limits. To counter this, other GaN crystal makers have constructed sophisticated autoclaves based on nickel super-alloys, which has come at a cost.

“These super-alloys have better high-temperature properties, including strength and creep resistance, but are much more expensive and difficult to scale up,” he says. “In addition, the crystal growth rates [within these autoclaves] are quite low.”

To tackle these problems, D'Evelyn and colleagues have developed a proprietary GaN crystal growth technique, called SCoRA - Scalable Compact Rapid Ammonothermal - that they believe will cut the costs associated with existing GaN ammonothermal crystal growth, and produce commercial-grade substrates. Instead of using a super-alloy autoclave, the team has constructed a robust steel reactor lined with an insulating ceramic layer to protect the steel from high, internal processing temperatures.

As D'Evelyn explains, the crystal growth components are placed within a compartment inside the two-layer pressure apparatus, which is filled with ammonia and then internally heated to kick-start crystal growth. “The outer metal is cooled externally, doesn't get very hot, so steel works fine for us, thank you very much,” he adds.

But has Soraa successfully grown high quality GaN crystals? Yes. A laboratory-scale reactor, built prior to the ARPA-E funding of early 2011, demonstrated the technique's feasibility. And with the government cash, the team has since built a pilot-scale reactor and grown crystals at, says D'Evelyn, higher growth rates than rival companies.

“[Competitors] have demonstrated two-inch crystals of extremely high quality,” asserts D'Evelyn. “We have grown two-inch crystals. There is a range in the quality of our crystals, with the very best crystals tending to be smaller and the largest ones not yet ready for manufacturing, but we are developing the growth process for high yield and high growth rates, and expect to produce extremely high quality four inch crystals in the future.”

While both D'Evelyn and his colleague, Mike Krames, chief technology officer at Soraa, assert these four-inch crystals will have a lower defect density than any HVPE-grown equivalent, neither will say exactly when these will be produced. “We have more work to do on ammonothermal growth and frankly, still have a lot more legs to go on HVPE substrates,” adds Krames.

Still, the future looks very bright for Soraa. Recent reviews of the firm's GaN on GaN LEDs cite a more vibrant and consistent beam compared to existing halogen lamps.

And excitingly, progress on their ammonothermal growth process will feed into the company's laser diode division, which is developing devices for displays and hand-held projector markets. “Using our technique, we have already demonstrated high quality [substrates] from the m-plane non-polar, semi-polar and c-plane polar planes of GaN crystals,” says D'Evelyn. “As our laser team explores a range of crystallographic planes we have told them that whatever they decide they need, we can make.”

 

Mark D'Evelyn, vice president of Bulk Technology, is confident Soraa's ammonothermal growth process will soon be producing high quality four inch GaN crystals.