PLYMOUTH, England In the first quarter of 2013 , Plessey Semiconductors Ltd. will begin production in Plymouth, England, of a range of high-brightness light emitting diodes (HB LEDs) based on gallium-nitride-on-silicon wafer technology.
The company said it is aiming at the solid-state lighting market and is already demonstrating devices and is working on the final product specifications in collaboration with potential customers ready for sampling.
It's a gamble for PLessey, which already has one MOCVD machine supplied by Aixtron SE (Herzogenrath, Germany) installed for the creation of the GaN epitaxial layer on top of 6-inch diameter silicon wafers. Essentially Plessey (Swindon, England) is committing the whole of its 6-inch clean room and potentially more space to riding the wave of consumer demand for solid-state lighting.
"We've got the customer requirements nailed down quite well. One has given us a very definite specification," said Barry Dennington, chief operating officer, in an interview with EE Times conducted at the Roborough manufacturing site.
Meanwhile the company is running a variety of its legacy IC processes on 200-mm diameter wafers in the other part of the cleanroom. These include CMOS, BiCMOS, BCD and silicon-germanium processes. It also produces image sensor ICs and the company's EPIC electric potential sensor technology there. Dennington reckons that the company is on an annual run-rate of $17 million or $18 million of annual sales out of the 200-mm fab, including some foundry work.
However, since the Plessey name was resurrected through the acquisition of former Plessey assets and intellectual property in May 2009 by Michael LeGoff, who serves the company as CEO, the company's avowed intent has been to become a component company.
The move into LEDs was prompted by support work that the Roborough factory had supplied to researchers at Cambridge University. Teams associated with Professor Sir Colin Humphreys had spent ten years and about $16 million in researching the layering of gallium nitride and including indium and aluminum on top of silicon substrates. CamGaN Ltd. was formed in 2010 to commercialize the technology and in February 2012 Plessey announced that it had acquired CamGaN for an undisclosed sum.
The ability to engineer out the strain of lattice mismatch between silicon and gallium nitride while tuning the light emission of the LEDs is a breakthrough according to Dennington.
A common alternative is to deposit GaN on sapphire or silicon carbide substrates but these typically only come in 2-inch or 4-inch diameter wafers. Achieving low defect densities with high levels of light extraction on larger wafers contributes to higher yield, while the ability to limit layer thickness reduces manufacturing time and material costs. GaN layer thicknesses on sapphire and silicon carbide substrates are typically 8 or 9 micron versus Plessey's 2.5-micron layers over silicon.
Dennington reckons Plessey can achieve cost reductions of 80 percent compared with LEDs made on silicon-carbide substrates and 50 percent compared with LEDs made on sapphire substrates. Partly that is because the substrate material is cheaper at just $20 per raw wafer and partly because it is bigger thereby helping with throughput and yield. "We think we're 18 months ahead of anybody else trying to produce high volume GaN on silicon wafers," said Dennington.
Click on image to enlarge.
The first blue LED samples are
characterized by peak emission at a wavelength of 460 nm with typical current
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for todays commercial processor giants such as Intel, ARM and Imagination Technologies.