Compound semiconductors point the way ahead for life after silicon
Image credit: CSA Catapult
Compound semiconductors don’t sound that exciting, but silicon’s physical limits are being reached and alternatives need to be developed. The UK is launching a facility that aims to put it at the forefront of this new technology.
Silicon is brilliant. It is the material we have harnessed to fuel the digital revolution. The vast majority of our technology depends on it and will continue to do so. However, it has limitations.
Firstly, it is getting too small to be stable. The latest technology node is 7nm, which is little more than marketing speak for ‘smaller than the technology before’, which happens to be 10nm. Even though transistor features are not actually 7nm, they are still getting too small to operate reliably. They would have to continue getting smaller in order to satisfy the famed Moore’s Law (double the transistor density every two years), and this does not seem possible with current technology.
Silicon also suffers from having relatively poor ‘electron mobility’, which essentially determines the speed at which transistors can act. Thirdly, it is not good at high temperatures and, finally, it is not the best at dealing with light. This last point is important as photonics is likely to play an important role in pushing forward computing power.
Assuming photons cannot carry the can on their own, new materials are therefore needed if the cutting edge of technology is to retain its sharpness. And this is where the UK’s Compound Semiconductor Applications (CSA) Catapult comes in, a facility that is now preparing to open its doors to the electronic world.
Compound semiconductors are combinations of two or more chemical elements, for example gallium nitride or gallium arsenide. The reason that they are of interest is they operate at much higher frequencies than silicon. They also operate at much higher temperatures and they emit and detect light much more efficiently. So they find applications in areas such as power electronics, RF communications such as Wi-Fi, and photonics such as lasers and solar cells.
Dr Andy Sellars is chief business development officer of the CSA Catapult. He explains that there are many varieties: “There’s indium phosphide, gallium nitride, gallium arsenide and then you can do strange things with them. If you have gallium aluminium arsenide and then vary the proportion of aluminium and arsenic, you can play very weird tunes with them.” And those ‘tunes’ could represent the technology breakthroughs that are the key to success for whole, albeit often niche, industry sectors.
A big part of the CSA Catapult’s role will be talking to the 5,000 or so end users of compound semiconductors in the UK and helping them develop a roadmap. Sellars says: “For example in 5G, we can talk to all the companies that operate in RF communications and we can ask ‘what is the roadmap? What do you want to achieve by 2025, and by 2030?’. Then we can work back into the supply chain and talk to the academics and say ‘what research can we pull forward?’. Then the Catapult could commission some chips early on in the supply chain; one part of the Catapult strategy is actually to produce development kits.”
If a chip is produced on its own then a whole circuit needs to be developed to test and evaluate it. Getting a development kit together removes this hurdle and allows end users to integrate the development kit into their systems and get designers designing with that chip. It is standard practice in the silicon world.
Sellars says: “Our part of the Catapult strategy is to do the same but for compound semiconductor chips. So we can talk to the end-user base and say in aerospace ‘what are your requirements?’, or in automotive, or marine, ‘what are your requirements?’. If there’s enough commonality, we will go back to the supply chain and say ‘can you provide us with, for example, the silicon carbide chip?’. We can chuck it on the board, give it a peripheral interface, then offer it to those companies and say ‘go and evaluate it’.
“If it works and you’re happy you’ve then got all these end users that will drive the sales volumes back to the fab. So we could commission these chips using UK fabs and that will drive the volumes up by finding multiple application areas. The more applications you get, the more volumes you get for the chip and then as the volumes of the chip go up the yield goes up and the cost of the chip comes down, it becomes more sustainable.”
It’s an exciting opportunity for the UK to confirm its leadership in this field and it is built around some serious investment in pure research. In fact, the Engineering and Physical Sciences Research Council (EPSRC) has ploughed around £750m into early stage research over the past decade, and has announced further investments including £80m in the Institute for Compound Semiconductors hosted at Cardiff University, a £10m investment in the Future Compound Semiconductor Manufacturing Hub at Cardiff University and the Compund Semiconductor Centre, a £40m joint venture between Cardiff University and IQE. It will be no surprise that the new CSA Catapult is also in Cardiff. Although the Welsh capital is the focal point, the expertise is scattered around the UK at universities including Sheffield, Glasgow, Imperial College and Warwick.
By filling the gap between this extensive research base and those 5,000 or so systems integrators, the Catapult completes a supply chain that is uniquely positioned. Sellars says: “The total investment in this area is around £180m – larger than other investments globally. But what we have here is something which spans the supply chain from wafer, die, device to system and it expands the research grade level from technology readiness levels 1 up to 6.” At which point we link to the commercialisation end of the chain.
Part of the challenge is to wean the industry off silicon. It has been researched to the nth degree and given an operating environment engineers can be absolutely sure about how their silicon will work. That is not the case for, say, a gallium nitride transistor – there is some uncertainty in it. Additionally that gallium nitride transistor needs an interface back to the silicon and that relationship needs to be reliable. Part of the role of the Catapult is to de-risk that.
“What will happen is that companies will take small steps down this route,” claims Sellars. “They won’t drop what they’re doing in silicon, but they know it’s inefficient, so they’ll start to replace parts of the system with gallium nitride or silicon carbide to get small improvements in efficiency. Small incremental steps. The role of the Catapult is to try to and make that go faster, try and make their money go further.”
One example where compound semiconductors could make a difference is in RF to digital, digital to RF conversion for space applications. In satellite communications there is an enormous amount of weight required for waveguides, and to put electronics into space costs about £30,000 per kilogram. So the lighter you make this the better. There’s also lot of radiation up there and silicon can be susceptible to ‘soft errors’ as a consequence, and so silicon needs to be screened. Compound semiconductors, on the other hand, can operate in a high-radiation environment. However in satellite communications a lot of the work is analogue. “One requirement is potentially a direct RF to digital, digital to RF conversion, like an ADC operating the RF,” says Sellars. “That would allow us to produce more software-controllable satellites, that would be lighter in weight. The reason why that’s important is satellite communication bandwidths are just going through the roof. And with cubesats, constellations of smaller, cheaper satellites, everything needs to be miniaturised and much more lightweight.”
‘There’s up to 2.5 million engines per year, 1.5 million cars per year - we could actually do the whole supply chain in the UK.’
Photonics is another very promising application area. In particular, tuneable lasers have a whole range of uses. One application is in quantum,where lasers are used to chill an atom down to a certain temperature, just above absolute zero, at which point it displays quantum behaviour. Very precisely tunable lasers are required for this and compound semiconductors have the optical performance to achieve it.
This same performance could also be useful for VCSEL – vertical cavity surface emitting lasers – which are relatively cheap to fabricate, and this results in numerous opportunities. Sellars comments: “One of the areas where VCSEL lasers might find new applications is in gesture recognition. Your future mobile phone, instead of using a touch screen that might use quite a rare material, might be able to do gesture recognition to let you control the device. That could open up new applications through VCSEL lasers.”
One of the markets with most potential is the migration towards electric vehicles, given the backdrop that in the UK we already make around 2.5 million petrol and diesel engines a year. With the drive towards EV technology there is the £246m Faraday Challenge to develop the necessary battery technology and this includes the power electronics to manage such features as regenerative breaking.
It is the epitome of the opportunity that the CSA Catapult is there to take advantage of, according to Sellars: “We’ve got hundreds of companies in the UK that can develop power electronic systems. We’ve got funding for the batteries. We’ve got an automotive manufacturing supply chain, some of it in the UK, not all. There’s up to 2.5 million engines per year, 1.5 million cars per year. We could actually do the whole supply chain in the UK – we could commission the power electronic devices through UK fabs then make the power electronic systems in the UK, put that power electronic system around the battery and couple that to the motor and make a whole unit in the UK. That’s quite a powerful linkage in the supply chain.”
Such a scenario represents a perfect outcome for the CSA Catapult, but it is not just wishful thinking. Although a fraction of the global $300bn silicon market, the compound semiconductor market is still substantial at $66bn. What is more, it is growing at three times the rate of the silicon industry with a CAGR of around 12 per cent taking it to $140bn by 2023. Most significantly, the UK already has 9 per cent of that global market thanks to its historic investment at the R&D level.
Sellars believes the CSA Catapult will provide the link to the rest of the supply chain to see this proportion grow.
Investing in innovation
“They are about market failure,” is the surprisingly downbeat comment from Andy Sellars, when surely the Catapult programme should be built on success? But apparently not, or at least not in a business sense – the technology could, and should, be leading edge. “If a market is growing at a good healthy rate than the government should allow that to happen, it shouldn’t step in,” explains Sellars. “Where Catapults operate is where things are just too risky for business to invest. At that point if you don’t intervene the danger is that that investment takes place somewhere else in the world.” The Catapults, and there are currently nine with the CSA Catapult about to be the tenth, are aimed at filling this void in specific areas where there is deemed to be most potential.
The criteria are that there is a large growing global market that we can tap into and that can grow as a consequence of central funding. There must also be the academic expertise that can be called on to provide the building blocks for development of both technology and market. Sellars says: “You wouldn’t put a Catapult into an area where the UK is lagging decades behind other countries. It wouldn’t make sense. You put it where the UK is in a pre-eminent position and we can accelerate that position.”
The nine existing Catapults are Satellite, Digital, Offshore Renewable Energy, High Value Manufacturing, Future Cities, Cell and Gene Therapy, Transport Systems, Energy Systems and Medicines Discovery.