VON ARDENNE: Beyond silicon carbide and gallium nitride – about gallium oxide and the advantages of PVD coating

Aalyia Shaukat (AS): Hello everyone. Welcome to today’s edition of “Power Corner”. I’m your host, Aalyia Shaukat, editor-in-chief of Power Electronics News. Today we have the pleasure of speaking with Guido Überreiter, VP of Semiconductor Strategy at Von Ardenne. Guido, how are you doing today?

Guido Überreiter (GÜ): Very well. Thank you very much. Thank you very much for the opportunity to speak here.

AS: Thank you for being with me. I appreciate it very much. I’d like to start with a brief introduction. I did a little research on Von Ardenne. As far as I know, it is a German company for vacuum coating systems which, despite having more than 1,000 systems in over 50 countries, is still a family business run by Pia Von Ardenne. And as far as I know, Von Ardenne focuses specifically on physical vapor deposition, or PVD, in the semiconductor sector.

The company manufactures coating equipment used in MEMS, piezo MEMS and advanced packaging technologies that integrate power semiconductors with logic chips, for example. And I’ve learned that researchers at Von Ardenne are actively working on processes for depositing thin films of gallium oxide (Ga₂O₃) specifically tailored to device manufacturing.

So maybe we can start with Ga₂O₃ as a semiconductor first. What are its advantages, and what makes it a compelling candidate for high-voltage power devices beyond 1 kV?

GÜ: Yes, I think one of the interesting aspects of wide-bandgap materials is that you are actually looking for larger and larger bandgaps to be able to switch higher and higher voltages. So today we have silicon carbide (SiC) and gallium nitride (GaN).

SiC works in a similar voltage range, but Ga₂O₃ offers several advantages that we can talk about later. However, the industry is looking for better and better power semiconductors, which I think is simply driven by the growing importance of energy conversion and efficient energy conversion. Initially, we talked a lot about SiC and GaN in the context of vehicle electrification. And in the last two or three years, with the boom of AI, suddenly everyone is also looking at the contributions of megawatt data centers.

And when you consider that 20-25% of energy is lost just in the path from power generation through all the distribution to the actual CPU, there is a huge energy saving potential if we make better and more efficient power semiconductors.

And Ga₂O₃ is one way to achieve this. Like any material, it has many advantages, but also some disadvantages that we can talk about. Overall, however, I think it’s a very interesting material that’s still at a relatively early stage, but holds promise from an efficiency and cost perspective.

AS: Yes. As you said, Ga₂O₃ has been researched for decades. What has changed in deposition technology that makes it viable for power device manufacturing today?

GÜ: Yes, I think it’s such a supply and demand dilemma sometimes, isn’t it? If you’re too early and there’s not enough demand, and people are happy with what they have, it doesn’t get off the ground.

But I think there are several things that have changed. For one thing, as I said, there is a lot of demand. Secondly, I think we have now developed methods of producing Ga₂O₃ that are much more reliable and efficient, so that the material is actually suitable for the production of power semiconductors.

In addition, I think thin-film technologies are now so advanced that you can also produce very detailed and very fine layers on the actual substrate – because it’s not just about the material itself, but about how you actually use it. And in my opinion, this is precisely where Von Ardenne’s strength lies: a variety of different materials and a variety of different processes for their deposition.

Here we see: The time is ripe to make greater use of PVD for this type of application too. And I think, as I said before, there is a huge demand – electric vehicles, grid conversion – all these things are coming together. And I think it’s also the point where there could be another small revolution in power semiconductors.

AS: Maybe we could talk a little bit about this substrate growth technology that Von Ardenne is working on and describe the advantages of Ga₂O₃ over SiC or GaN in terms of scalability of substrate growth.

GÜ: It’s true – when I joined Von Ardenne a few years ago, I was always amazed at how they use the materials. They’ve been using certain materials for years that are only now finding their way into the semiconductor industry – PVD is generally a very cost-effective way of depositing materials.

And that’s actually one of the disadvantages of SiC and GaN: it’s still very, very expensive to make these wafers. It is relatively difficult to produce them on large substrates. So as far as the transition to 300 mm is concerned – the industry has fought very, very hard to get SiC to 300 mm and we are slowly getting there worldwide. Still, it’s a very, very expensive approach, very time consuming and also very energy intensive.

And that’s where I think PVD technologies can make the difference – not just in depositing layers on a substrate, but one of the technologies we’re currently developing is to actually grow the material in the specific crystallographic structure using PVD technologies. So it’s not just about depositing layers; it’s actually also about building the substrate itself in a PVD process, which is quite novel, and it’s even more cost-effective because you can actually work with the same set of tools.

AS: So, just as an example – tell me if I’m missing something – I know that GaN needs to be grown on specific Si[111] wafers. Is it anything like that, or is that, as you said, not necessary? Can you customize the structure when growing Ga₂O₃?

GÜ: Our technology is basically able to create specific crystallographic structures atom layer by atom layer. So we have a tool that is actually capable of growing, and you can adjust the process configuration to get the desired physical parameters – and therefore also the electrical parameters.

If you look at GaN structures, in many cases it’s not just GaN. It’s GaN, then a bit of doping is added here, followed by another GaN layer and finally gallium doped with aluminum.

So there are many mixed layer stacks. The fundamental question is always: How do you build up the base layer? And we think we can do both.

While with GaN you basically need MOCVD-like processes to make the actual substrates, we think we actually have an advantage here – using an element that we’ve been working with for a long time and doing a few tricks on the PVD side to basically create the base layer, but also the overlying layer stacks that are required because of some of the shortcomings that Ga₂O₃ has as a material.

AS: Okay. And maybe we could talk a little bit more about the attractiveness of Ga₂O₃ for commercial use. According to my research, p-type doping in Ga₂O₃ has long been a challenge for full device integration – that is, creating the p-n structures needed for diodes and transistors. Does Von Ardenne’s deposition approach play a role in solving the p-type challenge? Or is that a separate problem that needs to be solved up front?

GÜ: Well, I think it’s actually the material itself and not the way you make the material. The fact that you can’t make p-type doping – or that it’s very difficult and actually very inefficient and therefore not desirable – is something that we’re not going to change no matter what method we use in making the material.

What is driving this, however, is a lot of innovation in terms of how you actually build devices on top of this layer – either with Schottky-type devices or with MOSFETs. The industry is then typically moving into the third dimension, building layers on top of it and constructing devices in a different way than just embedding them in the actual Ga₂O₃.

We’ve seen this with other approaches as well – GaN is very similar, not from a p-type doping perspective, but from a device architecture perspective. There are many vertical devices on GaN layers, and that’s exactly what we see with Ga₂O₃: overcoming this material engineering challenge through other device architectures.

AS: Interesting. Could you talk a little bit more about these alternative device architectures with Ga₂O₃?

GÜ: Well, if you want to build other types of devices, you basically start from scratch again by putting layers of other materials or differently doped materials on top. And what is then extremely important are the interfaces of these layers.

In the case of GaN, for example: How exactly do GaN and AlGaN interact at the interface and create a very clean surface with which you can tune the device parameters very well?

And this is exactly what we will see with Ga₂O₃: Components that are basically built vertically, where these transitions are on top of each other instead of next to each other. And I think that’s what the industry is increasingly using, also to optimize SiC and GaN. So the same methodology can also be applied to Ga₂O₃.

It will be important to design the surface very precisely, to deposit the next layers – which in turn will be very thin – very precisely and to obtain a clean surface and a clean interface. And with these alternating layer stacks, you can make devices that take advantage of the large bandgap and high electric field strengths of Ga₂O₃.

AS: Thank you very much. Let’s talk a little more about PVD techniques and their advantages. You mentioned magnetron sputtering or pulsed laser deposition. How do these processes compare to, for example, metal organic chemical vapor deposition (MOCVD) or hydride vapor phase epitaxy (HVPE) in terms of coating quality, scalability and cost?

GÜ: Well, I think in general PVD is very cost effective. It’s much cheaper and much faster. And if you think about some of our equipment, which is basically processing dozens of wafers in parallel as they’re spinning in the equipment and we’re depositing atomic layer by atomic layer at high speed, you can see how that creates an overall cost advantage in manufacturing those wafers.

And because we’re not limited by the wafer diameter, we can basically get as big as the substrates that we can grow things on. That’s actually another advantage that also has an impact on cost. The interfaces that I mentioned – producing clean surfaces that are very smooth and very homogeneous – is something that PVD clearly does better than CVD technologies or anything like that.

What we need to do – and this is the engineering challenge on our side – is to make sure that the crystal structure is exactly what we want it to be. I think that’s where the engineering excellence and material science comes into play with PVD, because for a long time that’s what has made applications more focused on these more expensive deposition technologies.

But we believe we have some very good engineers who know how to mix the concrete, so to speak. And that’s where I think PVD is going to play a much, much bigger role in power semiconductors because of the advances that engineers have made in materials and surfaces.

AS: That’s very interesting. This is a personal question – have you made power devices using the Ga₂O₃ technology you’re working on so far?

GÜ: No, we’re still at the stage where we’re depositing layers. We’ve actually been very successful in making what I would call alternating stacked layers – not moving one wafer to the next chamber and back again. We basically have a tool that just spins, and with each rotation it gets a certain coating of material. And as it spins, you can change the composition of those materials.

So you run the flywheel, and as it runs over time, you build up layer A, then layer B, then layer A again, then layer B, and you can mix them very well and very quickly. And that’s where we’re seeing a lot of interest at the moment – where people are saying, “Oh, I can actually make my components much faster with very different materials.”

AS: That’s very interesting. What application areas – what you can’t tell, you can’t tell – are showing the most interest in that?

GÜ: Well, like I said, Ga₂O₃ is the material that we’re focusing on, but those same tools are also basically capable of making GaN and, you know, mixtures of layer stacks, and that’s just the power electronics area.

You can also use them for MEMS applications – to grow certain layer stacks or materials into a very thick stack in a very short time. It’s basically the same tools, and we’re actually more advanced on the MEMS side.

But basically the same physical and chemical principles apply when you’re dealing with Ga₂O₃, GaN or any of these materials. And as you mentioned, the band gap keeps increasing – there’s boron nitride (BN) and then there’s diamond materials at the end. But it’s basically a similar kind of problem that you have to solve: getting the right crystal structure and growing it very homogeneously and very quickly.

And we believe that with the tools we have at our disposal, we can actually change the materials and therefore change the electrical and physical parameters of these materials as well.

AS: Fascinating. Anything else you’d like to add, or are there any questions I may have missed? You’ve already answered quite a few of my questions.

GÜ: Well, I guess there’s still the caveat, right? Where does Ga₂O₃ stand from a market launch perspective? Because right now it’s not yet mass production.

I think we’re still at a stage where a lot of the players in the power semiconductor space are ramping up their next-generation SiC fabs for mass production. We’re seeing great progress in GaN and different variants of GaN-based semiconductors.

I think Ga₂O₃ needs a few more years to get out of that experimental phase – and out of that phase where application engineers can actually make real devices from the materials – to something you would call prototyping or first product.

My personal guess is that it will be another five years before we see the first products, and then probably another 2-3 years before it goes into real mass production.

But honestly, I think we all assume that it will take us until 2030 to get to a $1 trillion industry. And now the industry is saying we’ll be at 1.3 trillion dollars by 2027. So it seems like a lot of things are accelerating, and I think in some ways that’s also a wake-up call for the traditional semiconductor manufacturers – because I think the SiC era is going to be over a lot sooner than a lot of people think.

AS: Well, thank you very much. I really appreciate your time, Guido. That was a fascinating conversation. I’m excited to see where Ga₂O₃ is going, what role PVD will play and how Von Ardenne will drive this technology forward.

GÜ: Perfect. That fascinates me too. Thank you.

AS: Thank you.

– – – – – –

Further links

👉 https://vonardenne.com  

Photo: VON ARDENNE