This quarter, Plessey Semiconductors plans to sample its GaN-on-Silicon LEDs. Samples are due out in December with production sales beginning in March for both LED die and packaged devices.
It is the culmination of a decade of research, development and commercialisation.
“When I was looking to acquire the Plymouth (Roborough) site (in 2009) I came across some skunkworks projects, being carried out under the noses of the German owners, doing experiments in depositing gallium nitride on silicon for LED,” Michael Le Goff, CEO of Plessey Semiconductors, tells me, “when I went round the site, the people doing this said the owners were not interested in it. I started looking at gallium nitride on silicon, and at LED, and decided this was a real hidden little gem.”
In Cambridge, the world’s greatest expert on GaN, Professor Sir Colin Humphreys FRS, had been leading the Cambridge University Centre for Gallium Nitride for ten years spending some $16m researching the material.
In 2010, the Centre’s expertise was spun off into a start-up company called CamGAN to commercialise the research in LED applications.
“In 2011 I made a proposal to a private equity group,” says Le Goff, “they did the research and decided to acquire CamGAN.”
“CamGAN’s process is unique,” says Le Goff, “unique in the world.” The unique value of the CamGAN process is that it uses ordinary silicon wafers as the manufacturing substrate, meaning that the products made on it are much cheaper than the usual sapphire-based or silicon carbide-based substrates normally used for making LEDs.
The Plessey LEDs are 80% cheaper to make than LEDs made on sapphire substrates and 50% cheaper than Silicon Carbide LEDs.
“The cost reduction comes from four factors,” says Le Goff, “first because the base material is cheaper than sapphire or SiC; secondly from the wafer size – we’re making them on six inch wafers which nobody else does; thirdly, because the substrate is silicon it reduces manufacturing time – sapphire and SiC require more handling and hand-carrying; fourth, because it runs on a standard IC line the yields are much better. We expect 90% yield; the yield on sapphire-based processes is 20-30%.”
The importance of cost to the adoption of LEDs can hardly be over-estimated. It is generally agreed that the LED market will take off when LEDs are adopted for general lighting applications.
At the moment, the $20 or so for an LED bulb is seen as too expensive for the domestic user. It is thought that, taking into account all LED’s other advantages, $10 is the point at which LED will enter the general lighting market.
So cost has to be reduced by 50%. And 50% is the scale of the cost reduction delivered by the CamGAN/Plessey process.
Each six inch wafer produces a potential 14,000 LEDs
Key to manufacturing LEDs is the MOCVD (Metal Organic Chemical vapour Deposition) process step. Plessey has installed an MOCVD machine at its Roborough fab. It is capable of an annual output of LEDs worth $20-25m.
Once demand looks like overwhelming the output of one machine, the plan is to follow the market.
“We have a plan to add three more MOCVDs to follow the sales path over the next 18 months,” says Le Goff, “we have the space at Roborough to have another eight MOCVDs.”
There are three possible ways forward, he explains.
“One there’s the sensible route. Build capacity when you have the cash. Avoid debt. Use vendor-financed tools.”
“Two, where there’s an abundance of these MOCVD machines, as in China, do a jv. The China government has been subsidising the manufacture of MOCVD machines by 80-90% and now there’s a huge surplus of MOCVD machines in China. It’s a huge opportunity. The problem with it is IP leakage – you’re going to have your people going off to set up their own plants in competition with you.”
“Three,” adds Le Goff, “get the first one up and running and then buy 30 MOCVDs and get production volume. For every eight MOCVDs you get a billion LEDs – high brightness big LEDs – if you have 30, that equals four billion LEDs a year and you can drive the market.”
It is expected that LEDs will be a 180bn unit market by 2015.
An MOCVD machines costs $2m. “We could get £50m and build a factory,” says Le Goff, “if we start getting constant demand, and quickly run into capacity issues, we need to be able to very quickly ramp up capacity.”
The Cambridge research people are continually improving the product from the materials side while the Plessey research people are improving it from the device engineering side.
The research roadmap is to deliver more efficient devices with greater light output and, eventually, more integrated devices with built in driver ICs and power management technology.
“That is looking at the next round of competitive advantage,” says Le Goff, “we have a roadmap of increasing performance.”
LEDs are not the only fledgling business at Plessey. The other is the sensor business. Plessey has a contactless sensor requiring no physical or resistive contact, which is being used for ECG measurement.
The sensors, called EPIC, work by measuring tiny changes in an electric field in a similar way to a magnetometer detecting changes in a magnetic field even at a distance and through clothing.
A feature of the technique is that it can be used for security motion sensors and non-contact electrical switches as well as for medical applications detecting heart beats, nerve and muscle activity either by dry contact (i.e. no gels required) or contactless.
Plessey has cost-reduced EPIC via a new electrode design to replace the expensive titanium dioxide electrodes currently used for medical applications. The sensors now sell for a dollar apiece in volume quantities.
Using EPIC sensors, Plessey has produced an ECG machine for a production cost of $50. Plessey calls the machine ‘Impulse’ and it is currently with the US FDA for approvals.
“The original idea was to provide a demonstration to show customers,” says Le Goff, “then we felt if we can get some sales then we could say to customers; ‘Look we can get sales so you should be able to get sales’.”
Plessey has also put the sensors into a pad to go behind a car driver’s back to detect from the ECG results whether the driver is getting drowsy and warn him or her with an alarm. Or they have been put in a mattress to provide an ECG reading as soon as you get into bed. And they have been put in wristbands to measure heartbeat for people taking exercise.
Asked if it is Plessey’s intention to move from the components business into end products, Le Goff replies: “In hushed tones – we would like to sell end products. We have our eye on it. We’re aware of it if we’re going to become a consumer electronics manufacturer. It’s in our plans.”
He has no shortage of ambition. “We want to build Plessey to be a market leading player employing thousands of people,” says Le Goff, “I want Plessey to be a global electronics company. We want to get it back to where it was.”