Why silicon still matters
Silicon IGBTs in the spotlight: Thomas Hauer, Avnet Silica, explains why the enduring workhorse of power electronics isn't going anywhere.
BY REBECCA POOL, TECHNOLOGY EDITOR
What applications are driving demand for silicon IGBTs right now?
Hauer: Both induction heating and heat pumps are the main drivers – and with heat pumps, it's getting interesting. Suppliers specify plus 20-year lifetimes, and with IGBTs, the [long-term] effects of switching on the motor winding to winding insulation are understood. But silicon carbide [devices] switch faster, and so the insulation is under much more stress. Many suppliers are left thinking, 'well, we cannot really [harness] the positive effects of silicon carbide because we need to make it switch slower, like the IGBT.' So why should they implement a more expensive silicon carbide MOSFET, which only makes life more complex, instead of just sticking with the silicon IGBT?
Where else might a supplier prefer silicon IGBTs over SiC MOSFETs?
Hauer: This is a little dependent on how the module is designed, but silicon is more resilient when it comes to short-circuit ruggedness and is usually more rugged in terms of overload capabilities – this is why the industrial motor inverter can be designed to last more than 20 years or so. I know there are all the bells and whistles and marketing going into wide bandgap semiconductors at the moment, but silicon still has its space, and it will stay there. I'm not going to advocate a customer towards silicon carbide, if they don't want this - if the application makes sense in silicon, why would you waste your time investigating wideband gap alternatives?
Are there applications where silicon persists despite expectations of a switch to SiC?
Hauer: Personally, for me, it's the industrial motor sector. We had a big, big push from customers who wanted silicon carbide, but then they faced obstacles, with the output filters, the pricing, the wiring and more. They finally decided, 'well, why should we do this?' Silicon carbide-based inverters are really good if you are integrating them on top of the motor, say, but in most industrial robots, the inverter sits close to fabrication, next to the robot arm – so we've seen customers trying to figure out if there are any benefits to silicon carbide here. I recently had a discussion with a customer that wanted a five-amp DIN-rail power supply, and they asked, 'why on Earth should I use a silicon carbide MOSFET that can do 50 amps for absolutely no reason?' I think customers are now well educated and understand what silicon carbide can bring and what it can't bring. Silicon has stayed [in this application], and will stay here way longer than I thought it would.
Have you seen customer requirements for IGBTs shifting in the last 12 to 24 months?
Hauer: On performance, customers are increasingly looking into topside cooling – this started with silicon carbide in automotive drives but now the first IGBT is also available for topside cooling. We need to see if this application properly materializes for IGBTs and I have not yet made up my mind if this actually provides any benefits to the design – but topside cooling is nice, once you find a way to [apply] it.
Are you seeing any concerns over security of supply?
Hauer: Global politics are making customers nervous. As a distributor, we spent the last 15 years running around and asking customers to have a second supply source – many dropped this as it required a lot of effort in design, qualification and certification. Now, customers are very cautious. We have customers based in Europe and their customer base is European – while some are pushing purely for European suppliers, others want to have access to suppliers from other regions of the world but with a first, second, and even third source of supply.
What key hardware developments do you see taking place right now?
Hauer: A current trend in PV inverters is getting high input voltages and high input currents into bigger modules. The tricky thing here is that if you increase the input voltage, you also increase the link voltage, which means the B6 bridge output, [a three-phase inverter bridge with six switching devices], needs to run at lower duty cycles - so the losses you save at the input stage, you then gain on the output stage. In my point of view, this is the kind of thing that needs to be worked on.
Other hardware developments?
Hauer: In the four-wheel-drive of electric vehicles, we see silicon IGBTs being used in the boost motor [on the front axle] of these vehicles whilst silicon carbide MOSFETs are mostly used in the inverters on the rear [axle] inverter. Suppliers are now thinking of combining the silicon carbide MOSFETs and silicon IGBTs. It makes sense, because you want high efficiency from your automotive drive-train inverters across the entire [operating range] of the motor, not just at peak power. I'm an electric car driver myself, and use my first electric car as an example: peak power is 210 kW but average power consumption is roughly 19 kWh per 100 km. This means the inverter runs, on average, below 20 kW, and achieving high efficiency here is a big challenge for automotive OEMS. But you can use the silicon carbides for fast switching and then use the IGBTs for the heavy lifting - and the big advantage here is the IGBT is still cheaper than silicon carbide, which makes this a tempting 'thinking exercise' for lots of automotive customers. Still - there is a missing piece of the puzzle: IGBTs and SiC MOSFETs have very different gate drive conditions, and you cannot combine these two devices with simple gate drivers. And as I understand it, this gate driver problem isn't yet solved.
Where are opportunities for silicon in the coming years?
Hauer: Heat pumps are a steady business that's here to stay as we see this huge push to get away from gas and coal heating. In the DACH region [Germany, Austria, Switzerland] almost all of the newly built houses have heat pumps – so there is a huge business opportunity here. But I see opportunities everywhere where there is imminent price pressure and where efficiency is not the [metric] that is monitored the most – if you can spare, say, half a percent in efficiency then fine, go for the cheap option. One heat pump customer recently moved from wide bandgap semiconductors to silicon again, and then another customer decided to stay with IGBTs for their lower-end motor inverter.
Any silicon MOSFET innovations?
Hauer: Silicon MOSFETs are more about incremental innovation, and optimising certain figures of merit. I recently asked one supplier a semi-provocative question: you now have two by two millimetre MOSFETs, so when do we see one by one [devices]? This was a joke - two by two is already small. But we see silicon MOSFETs in top-cooling (TOLT) packages and also HiperFET Q3 package; the package variety in silicon MOSFET is enormously huge. The MOSFET market is under pressure because of global discounts, but silicon MOSFETs will continue to have their place in many applications where they are just good enough.
What about power management ICs?
Hauer: Last year, at ECCE Europe, a big discussion was the vertical integration of PMIC processors. Technically, it makes perfect sense – you need to have a stable voltage close to the processor. But thermally, it's a nightmare because on the bottom, the processor is producing losses and heat, and then you have the same again on the top - so cooling will be, for sure, a challenge. I do think that this is a problem that will be solved quite soon, because a lot of groups are working on this – but I'm interested to see how they will manage to solve this. When you look at the gigawatt hours of energy that run into a data centre - well that's a lot of cooling that needs to be applied.
What packaging innovations are you excited about?
Hauer: For discretes, I think the most famous package is the HU3PAK (made by ST Microelectronics). Development was triggered by Tesla, and the package was originally designed for 650 V silicon carbide devices in onboard charging - ST was the first to publicly release these devices to the market. Around 80% of the silicon carbide lines now have this package, but it's also being used for IGBTs. For power modules, injection-moulded dual-side [substrates] are interesting for IGBTs. So here, you basically have a sandwich construction, where the modules sit in between some aluminium blocks, through which cooling liquid flows.
What does the future hold for silicon devices?
Hauer: I see suppliers working on a new generation of IGBTs – and this is a good thing. Budgets are not as big as wide bandgap semiconductor budgets, but as long as suppliers continue to invest in a technology, that's a good sign. I had a conversation with an analyst about IGBTs about growth numbers, and while the figures for silicon carbide and gallium nitride are, of course, shiny, we see steady growth in IGBTs, and I don't think we will see this going away.
Key takeaways from Thomas Hauer
• Silicon IGBTs remain relevant in applications requiring ruggedness, long life, and cost.
• SiC offers fast switching and efficiency gains but adds stress to motors and complexity in design.
• Silicon is being used with SiC packaging innovations, such as HU3PAK, to enhance thermal management.
• Hybrid designs (IGBT and SiC) are being explored but require advanced gate drive solutions.
• Heat pumps and industrial inverters represent steady growth opportunities for silicon devices.
Thomas Hauer is a Power Technology Specialist Manager at Avnet Silica
