Cree is investing this huge amount of money into expanding capacity. Will the spending on research and development also increase?
Gregg Lowe: We would be very foolish, if we didn’t. Our R&D expenses are fixed at ten to eleven percent of our annual revenue. So as revenues increase, we will also raise our R&D expenses accordingly.
At the moment, we are hiring people at a very high pace. And when they come in, they see that we are in a leadership position, and with this announcement today they see that we are absolutely committed to being successful. The focus of the R&D team is to drive the on-resistance down as well as cost and to increase the yield. And by further reducing costs, other applications can switch from silicon to silicon carbide.
How many people are now working for Wolfspeed?
Gregg Lowe: Our total workforce has been relatively stable, but the percentage of employees focused on Wolfspeed has increased, and we plan to continue hiring R&D, sales, marketing and applications engineering roles to help to drive Wolfspeed growth. We are still in take-off mode and working for Cree and Wolfspeed is considered very attractive!
Let’s turn to products. How will Wolfspeed evolve its device portfolio over the next years?
Gregg Lowe: Today we have announced two new 1,200-volt chips targeted for inverters in electric vehicles, 13 milliohm and a 16 milliohm devices, both as bare dies intended to be used in power modules. They are manufactured in our third-generation technology on 150 millimeter wafers.
So they are still planar MOSFETs, aren’t they?
Gregg Lowe: You are right, they are planar. Over time we will move to a fourth-generation technology which could be planar or could be trench. We have demonstrated both. But it’s a matter of fact that these bare die products announced here at PCIM have a lower on-resistance as any trench SiC MOSFET on the market today. We don’t care about whether it’s planar or trench, we care about on-resistance.
And planar is much more robust as trench, isn’t it?
Gregg Lowe: Correct.
I know that Wolfspeed is experimenting with silicon carbide IGBTs. Is there any news on that? When will they be available as a product? And what about Superjunction MOSFETs in SiC?
Gregg Lowe: For the market we are currently focusing on, the market for automotive inverters and charging stations, SiC IGBTs and Superjunction MOSFETs make no sense. For lower voltage devices – 1,200 and 1,700 volt – the drain resistance is not the dominant contributor to the RDS(on). Therefore, a Superjunction MOSFET won’t give us much advantage in this regard, but it’s very hard to manufacture such a device.
I see Superjunction MOSFETs and IGBTs in silicon carbide in the medium-voltage market with several kilovolts of blocking voltage. We’ve demonstrated SiC MOSFETs from 3.3 to 10 kilovolts as well as SiC IGBTs at 16 and 27 kilovolts. In the silicon world you switch from a unipolar device like a MOSFET to a bipolar device like an IGBT at 900 to 1,000 volts. As the electrical breakdown field of silicon carbide is ten times higher than silicon, the device architecture transition is also at close to ten times higher voltages – at some ten kilovolt. I can imagine such devices in grid-tied applications like solid-state transformers or very large electric drives.
Will we see silicon carbide IGBT and Superjunction MOSFETs products in the foreseeable future?
Gregg Lowe: I don’t think so. At the moment, we are completely focused on the market for electric vehicles, specifically main inverters and on-board chargers.
So Cree focuses on battery-powered electric vehicles. But there are many alternatives to conventional internal combustion engines, such as natural gas, hydrogen and fuel cells. Isn't it a little risky to rely only on battery-powered electric vehicles?
Gregg Lowe: In our wildest assumptions we expect a market penetration of 30 percent for battery-powered electric vehicles in the 2030s. Nevertheless, today this very market, as small as it is, is driving our business already. If you assume a market penetration of only 20 percent, it’s still a multi-billion-dollar market.
And by the way, if you take for instance a fuel cell, you have to convert that electrical power anyway to power the wheels and the car electronics. With our silicon carbide switches we can make that conversion much more efficient.
I’m asking, because many analysts foretell that automotive will trigger the success of silicon carbide; otherwise it will remain in the niche.
Gregg Lowe: I don’t agree. We are getting a lot of interest also from the industrial market. But if battery-powered electric vehicles really get that market share, it will be by far the biggest market for silicon carbide. That’s for sure.
OK, the issue with capacity should be solved with this one billion dollar investment. But what does this mean for device costs?
Gregg Lowe: Well, let it put me this way: For customers who purchase raw wafers from us, we have two options: Firstly, we can do the very normal business with them or, secondly, we can enter into long-term supply arrangements as we have done for example with Infineon and STMicroelectronics. Those customers, who have these long-term arrangements with us, get two major benefits from this: Firstly, they get a stable supply with raw wafers, and secondly, they get the full benefit of the cost reduction as our capacity grows.
Silicon carbide is now on its tipping point, so there is a lot of room for cost reduction and therefore for price reduction. Silicon was at that point in the 1980s. But today, the silicon supply chain is very mature, so there is little room for cost reduction.
But silicon carbide wafers will never reach the same price as silicon wafers, because they are grown energy-intensively from the vapor phase, will they?
Gregg Lowe: You are perfectly right.
But the price per ampere might get lower as with silicon devices, might it not?
Gregg Lowe: Now you have touched one key point. Since silicon carbide chips can handle much more amperes per square inch, the cost – and therefore the price – per ampere might eventually drop below the value of silicon MOSFETs and IGBTs.
Gregg, many thanks for taking the time.
This interview was conducted by Ralf Higgelke.