Q&A with Plessey
Myles Blake is the business information director for Plessey, a maker of advanced optoelectronic technology based in Plymouth, UK. He has more than 30 years of experience in the semiconductor industry, having joined Plessey Semiconductors in 1981 as a test engineer. Since then, Blake has helped guide the company through its IT processes, eventually moving from software development to project management.
Conducted by Jenny Donelan
Plessey, which used to be called Plessey Semiconductors, has a long history. Could you give us a synopsis?
The history, as far as the electronics side of it goes, dates back to around 1926. Plessey was involved in the Second World War, including the radar systems that came out at the tail-end of WWII. Plessey’s electronics were in the Spitfire [fighter aircraft]. So, there is a lot of British history there, but as far as semiconductors are concerned, that all started in 1956.
ID: Plessey has gone through a lot of changes over the years, including being acquired by GEC, Mitel, and then the German foundry firm X-FAB. When did your division of the company in Plymouth start on its current path?
MB: Right around 2009, a number of employees undertook a management buyout [from X-FAB], rebranded the company to Plessey Semiconductors, and began to focus on our core skills, which were largely around the sensor side of things. Quite rapidly after that, we acquired a specific technology from Sussex University for digital sensors, and later we acquired other capabilities from a Cambridge University spin-off. Then we put all of our focus into LED manufacturing.
The starting point for where we are now is our gallium-nitride-on-silicon (GaN-on-silicon) platform. This is the foundation technology onwhich we’re building everything else. We have a number of “high-bright” LED products for solid-state lighting. However, I think it’s fair to saythat the competition in that general area is quite fierce. Yes, we have some key benefits, including lumens-per-watt capabilities, uniformity andour ability to integrate traditional electronics components. But this vertical segment has become a “me too” market, and on the commercial side, itis quite, quite difficult to attract people to our product as opposed to paying a far lower price for something out of China.
So our focus today is around microLEDs, which is a technology that is also based on GaN-on-silicon.
ID: How long has the company had this microLED focus?
MB: The microLED focus started for us around the tail-end of last year, with a project to produce a microLED printhead, although I think Iwould use the term miniLED.
ID: Those distinctions are somewhat fluid right now.
MB: Yes indeed. We would say that if the pixel is less than 50 microns, it’s a microLED. Anything larger than that but less than maybe 100would be a miniLED.
But I need to step back to the GaN-on-silicon platform, which is core to all of our technologies, especially microLEDs. There are huge benefits to GaN-on-silicon, not the least of which is the cost of the substrate – the silicon itself. This is obviously far lower than sapphire substrate, and a great deal lower again than silicon carbide. (Generally speaking, I’m doing a comparison with sapphire, the other main substrate that’s on the market today.) And there are also scale benefits. For example, we can manufacture gallium nitride on 8-inch wafers. Admittedly, today our standard capability is around 6 inches, but we are rapidly moving down the 8-inch capability route, and have proved that we are able to manufacture on 8-inch silicon wafers. And of course the time will come when we will move to 12 inches. That’s a bit of a way off. It’s only recently that sapphire has started to move from 4-inch to 6-inch wafers.
As for other benefits, silicon itself gives you better thermal performance than sapphire, which means you can lower operating temperatures and therefore achieve higher reliability. As far as the light-emitting surface is concerned, the GaN-on-silicon essentially has a mirror effect. It’s actually putting the light where you want the light to be – out of the top of the die, not sort of bouncing through the die and substrate. This gives you an advantage in terms of any additional optical systems you need to add, because you’re simplifying those optical systems. Yet another consideration is around our historical capabilities as a company, which include incorporating additional electronic and optical components at the wafer level. That is one of the key application benefits that we have in terms of microLED capabilities. So the GaN-on-silicon platform is essential to where we are heading in the microLED space.
ID: What applications are you targeting in terms of microLEDs?
MB: We are of course talking about augmented-reality and mixed-reality in smartglasses, and also about pico projectors, smartwatches, mobilephone displays, head-up displays, and TV displays. Displays in general, to be fair. There are a lot of areas to target with microLEDs.
Today we are focusing on two products – a microLED illuminator and a microLED display. The first targets the incumbent solutions for, say, smartglasses and pico projectors. We are looking at replacing a light-source solution that is made up of three separate colors – the red, green, and blue – and a whole complex microsystem of lenses and prisms and mirrors – with a single-solution light source that is placed directly in front of the fly’s eye lens through to the LCOS or DMD.
This first product is almost like a step one. Step two is the production of our monolithic array solution. At this point in time when you read about microLEDs, I would argue that they’re usually not microLEDs, but miniLEDs. They are manufactured through a pick-and-place mechanism, which means the manufacturer has to pick up, not individually, but in a cluster, red dice, then place them, then pick up blue dice and place them, then pick up green dice and place them, and repeat the process until the array is full.
ID: That process is a big stumbling block for microLED manufacturing right now, isn’t it?
MB: Yes, absolutely. Whereas with our monolithic approach, you can, theoretically, within the bounds of your piece of silicon, produce a full array in one process. We are today producing pixels at the size of 8 microns. In terms of an RGB pixel array, that would be a 16-micron colored pixel, which would be structured from four 8-micron subpixels.
We have a way to go with this process. Today our array size is 0.7 inches on the diameter. From a monochrome point of view, this consists of 1,920 × 1,080 pixels, and the whole array runs at 60 frames per second. That’s 1,920 × 1080 pixels at the subpixel level, so if you apply that to the RGB, you’re looking at 960 × 540. Our intention is to have the monochrome array ready for CES in January 2019.
From there, we will move on to the second generation, probably in Q3 of 2019, which will be an RGB array that is 1,920 × 1,080 with 11.6-micron color pixels. It will have a slightly different structure, but that’s an aside, really. This will be around half an inch on the diameter. I can’t stress enough that this is a monolithic approach. We are starting with the 0.7 on the diameter and moving on to 0.5 inches on the diameter, full-RGB arrays.
Let’s not forget that we have to use a backplane to actually control the individual pixels. So today we have a silicon backplane that we acquire from a third party, which is Jasper [see this issue’s Industry News], and we have a chemical-mechanical polishing [CMP] process and a bonding line. We are in partnerships with other manufacturers to give us those capabilities.
The accuracy requirement for the bonding is phenomenal. Essentially we’re taking two separate wafers and bonding them together; everything has to align absolutely perfectly. With the CMP you have to have a very flat surface in order to be able to bond those two things together. We recently entered into a partnership with AIXTRON [see this issue’s Industry News] for the metal-organic chemical-vapor-deposition [MOCVD] reactors to give us the capability, accuracy, and consistency to manufacture all these microLEDs.
It’s very exciting. We think the GaN-on-silicon and monolithic microLED process can be applied to the whole display market. Of course, there is no need with today’s designs to have such small arrays in a TV. Samsung and LG are going the other route – these massive great display walls.
ID: What is the company’s biggest challenge thus far?
MB: Most of the biggest challenges involved the design concept – how could we make this work? I’m not saying we don’t still have problems to solve. We do. But the majority have been solved. Our biggest challenge going forward is how we can actually scale to manufacture and provide product into what is predicted to be a $5 billion market by 2023. How can we get all the partners together, and the supply chain, in a way that allows us to do this at a price that the consumer is going to accept? That’s the biggest challenge.
ID: What products are you actually in the business of selling right now with regard to LEDs? Are you in the licensing business as well?
MB: We are now selling the light sources, replacements for existing solutions in a single-source small form factor to go in incumbent projectors and some form of smartglasses. Going forward we will be selling the full monolithic RGB emissive array, which, again, can sit in smartwatches, mobile phones, smartglasses, pico projectors and so forth.
Today we are not in the licensing business. But we’re very proud of the fact that all of the key players in the market space are coming here to Plymouth to see our technology. As you know, we’ve got Vuzix [a maker of smartglasses and augmented-reality technology] and Huawei as partners, and there are quite a few others coming in from a customer perspective that I can’t share right now.
This 6-in. wafer has multiple 1,920 × 1,080 pixel arrays.
ID: In terms of scalability, are you looking at manufacturing on a global basis? You
can’t do everything in Plymouth, I assume.
MB: Well we’d love to think we could, and there is a degree of expansion that we can achieve within the four walls that we already have in Plymouth, and by that I mean we can expand to around 14 MOCVD reactors. And we would like to do that here in Plymouth. But 14 MOCVD reactors will be a drop in the ocean for microLEDs in five years’ time. So we will have to scale. Whether we do that through partners, through licensing, is yet to be determined. But we are fully aware of the fact that we have to scale and we are planning that as much as we can right now. It will be upon us before we know it.
ID: You’re an old company, and yet you’re a startup with this new technology. Can you tell us a little about your experience in those roles?
MB: Well, there have certainly been occasions when we have been too big for our boots. We come from that background of being a large, well-established business, and we expect all of our suppliers and customers to treat us like that, and we behave as if we’re a big business. That is a habit we can’t seem to get out of.
Another challenge, which is probably the biggest of all, and one common to every startup, is attracting investment, especially in the very early days. Yet another, and we are feeling it very strongly right now, is attracting the resources with the right skills. Part of the problem is our geographic location within the UK, a blessing and a curse. The specific skills that we need with this new technology that we’ve developed, the likes of optical engineers at this level, have been very hard to come by. We have to attract resources on a global basis because they are few and far between.
ID: So Plymouth is not a number one destination for engineering talent?
MB: It is a beautiful part of the UK, and tourism is probably the number one industry here. It’s a great place to live but it is far from anywhere else. Things are a lot better now – we used to have problems with the infrastructure, such as the internet and the
roads, and so on. That is all improving, but it’s still not the best.
We have a particular highly skilled mix of people here in this building in Plymouth, and we can be grateful for that, because they have the right characteristics to make all of this technology magic happen. They’re sitting down to design, develop, invent, and create things nobody ever thought of before.
ID: As an individual, you have been with Plessey for quite a while and changed focus completely over that time. Is this something that comes naturally to you?
MB: To me, yes. When I joined the company I was a traditional sort of test engineer. I rapidly moved into IT; not fixing PCs, but designing processes and developing solutions for the business. That was a previous role but I am still IT director. In this kind of a role, you need to learn new technology and develop an understanding of the foundation of the business in order to provide solutions for the business. The step from there to designing new technologies is a small one. So personally, I find it very easy and love the challenges on a day-to-day basis. •
This article is based on phone and email interviews conducted by Jenny Donelan, editor in chief of Information Display.