Driving commercialisation of gallium oxide
A European supplier of gallium oxide epiwafers is playing a pivotal role in the birth of an ultra-wide bandgap power electronics industry.
TWO CO-FOUNDERS OF NEXTGO EPI TALK TO EDITOR OF CS MAGAZINE, RICHARD STEVENSON
RS: Andreas Fiedler, as a co-founder of NextGO Epi, and someone who has been working on this material for a decade, you have been championing the promise of gallium oxide in power electronics for many years. Where do you expect this oxide to make the biggest impact?
AF: Right now, in solid-state transformers. The more renewable energies in our grid, the more we have to deal with smart grids and routing our grids, because of fluctuations from renewable energies. So far, this has been covered by big wheels in steam-based power plants, but since we want to get rid of those, because they typically produce steam by burning something we don’t want to have in our atmosphere, we need solid-state transformers to handle our grid. You need a semiconductor that can help you manage above 10 kV at a high-power level. Gallium oxide has the potential to fulfil that need.
There’s not enough material on the market yet. The research community was basically supplying material for itself, to get research done. Now big players want to jump in. Someone has to fulfil the need, and we can do this. We have to build a gallium oxide sub-supply chain, just for Europe, to be resilient for whatever will come.
The Aixtron CCS system with which NextGO Epi produces its Ga2O3 epiwafers.
RS: Ta-Shun Chou, as CEO of NextGO Epi, you have your
finger on the pulse of the gallium oxide industry. How would you describe the
current state-of-play?
T-S. C: From 2022, multiple suppliers started supplying at least 4-inch gallium oxide. Now more start-ups and companies are showing interest, from China, Japan, Taiwan and the USA. This year we see 6-inch and even 8-inch gallium oxide demonstrated. We believe that we are now in the transition phase. Industrial players are prototyping, showing that gallium oxide can be a competitor to silicon carbide and gallium nitride, rather than a cool material in published papers. Most likely, in the next 2 or 3 years, you will see the first gallium oxide product come out.
RS: Why is now the right time to have the emergence of NextGO Epi, a gallium oxide epiwafer supplier in Europe?
T-S. C: First, for semiconductor material, size matters. Now we see more 4-inch and 6-inch substrates available, and the price is coming down.
Our customers and partners are saying, ‘I want to try gallium oxide. May I have some epiwafers?’ This indicates that there’s a bottleneck in getting high-quality, large-diameter epiwafers.
Our partner, Aixtron, has a gallium oxide MOCVD tool on the market.
All three factors encouraged us to push the NextGO Epi further.
For epitaxial businesses, this business usually only exists at the beginning of a new material generation. If you come too late, there’s no place for you. So, that’s why we decided to join right now.
RS: You claim that NextGO Epi is the only European supplier of gallium oxide epitaxial wafers in Europe. Are there other epitaxial suppliers in the US and Asia?
T-S. C: There are several other epiwafer suppliers. They are in China, Japan, and the United States.
There are mainly three technologies for gallium oxide epitaxy: HVPE, MBE, and MOCVD. Our partners in the USA and Japan are more famous for MBE and HVPE. Some suppliers in China are also exploring MOCVD and MBE.
We believe that MOCVD, in general, is much better for large-scale, high-homogeneity epiwafer production. We have more than 10 years in MOCVD process development. We have a unique position in this whole ecosystem.
A.F: At this early stage of the transition of the material into the power electronics market, it’s good to have multiple suppliers. Most R&D departments don’t want a monopoly on materials. They say we wouldn’t invest in this if you were the only firm. If something happens to you, we need someone else to provide the material; otherwise, we will lose all the money we spent on development.
T-S. C. We have signed strategic co-operation agreements with six substrate suppliers. They are in China, Japan, and the United States. In general, gallium oxide is still too young to mention competition. Everyone wants to make the cake bigger. We are also happy to have multiple suppliers, with good quality and lower cost. All substrate suppliers want to send their substrates to us. If they provide a good substrate, it’s good for their reputation.
RS: Recently, restrictions in international trade have held back the compound semiconductor industry. How is this impacting the nascent gallium oxide industry?
The Hall measurement system to evaluate the electron mobility in the epilayer, a criteria for its high quality.
T-S. C. We foresee that getting the cost down for gallium
oxide may not be as fast as we expected.
The second point is that NextGO Epi is a Germany-based company. We are in a very neutral position compared with companies in, say, the United States or Japan. We are not restricted or limited by strict export controls.
RS: NextGO Epi has now been operating for over a year. But the history of the team goes back much further. How did the co-founders get together, and what provided the spark to transform three researchers into co-founding entrepreneurs?
A.F: We met during our studies at the Leibniz Institute for Crystal Growth (IKZ), where we developed the whole technology. I started in 2015, doing my PhD on this material system, overcoming issues to obtain high-quality material.
Andreas Popp joined in 2017, first as a postdoc. He did so well that in 2019 he became group leader for epitaxial growth.
His first hire was Ta-Shun, as a PhD student, in 2020. Ta-Shun finished his PhD in record time. He published the most papers, so far, for any PhD student, and wrote the most patents. I’m very happy to work with him because he’s not just smart – he’s also working hard.
We saw the clear potential of the material. In 2019, we talked to industries about gallium oxide. They were like: ‘We just invested in silicon carbide. What is gallium oxide? Why are you bothering us?’ Five years later, in 2024, industry came to us. ‘Hey, you were mentioning this gallium oxide. Do you have some samples for us that we can test?’
With this momentum switch, we can push our research into a real application. This is also the main motivation for the entrepreneurship. We did great research, and those papers are well received. But when I talk to people outside the research bubble, they never understand what I was doing. When I can say, one day, this application, what you hold in your hand, comes from my research, this is a very strong motivation to go in this direction.
RS: What challenges have you overcome to raise capital?
T-S. C. Usually, you cannot expect the investor to invest in you so you can buy hardware equipment. Although you are a startup, they expect you to have all the equipment to run the business. Luckily, we are able to solve this problem. We are linked very deeply with our host institute, so we are able to lease equipment, including MOCVD, from our host institute.
A second challenge is that, in general, a semiconductor business model is not very common in Europe, including the UK. So, it’s very unlikely you can find an investor with a background in evaluating semiconductors.
We are very lucky that we found an investor. They don’t know our technology, but they are willing to sit down with us, listen to us, and try to understand how good our technology is.
We also have very good support from the previous COO of Aixtron, Dr. Jochen Linck. He happened to be a Limited Partner of one of the VCs. He knows our technology because Aixtron is the vendor of our MOCVD tool. He does not just provide business understanding. He guides us and teaches us what an epi-company should look like, based on his experience. He’s the mentor of business for the founding team.
RS: NextGO Epi produces gallium oxide epiwafers using MOCVD. It’s also possible to grow gallium oxide by MBE, HVPE, and mist CVD. What do you see as the pros and cons of all these techniques?
A.F: MOCVD is the most mass-production-compatible technique. MBE, HVPE, and mist CVD suffer from either low throughput or single-wafer growth runs.
MBE is not just a single-wafer tool, and you have unmatched precision and purity in an ultra-high-vacuum chamber. But growth rates are very, very low for gallium oxide.
With HVPE, growth rates are very, very high, but samples are very rough, inhomogeneous, and have issues with structural defects.
Mist epitaxy is for the alpha phase. There are scientific reasons why most people focus on the thermodynamically stable beta phase. As soon as you process devices and want to ion implant or anneal your contacts, your alpha phase may form the beta phase, because that’s the thermodynamically stable one.
With the beta phase of gallium oxide, the first devices were made with gallium nitride processing recipes. They work like a charm. They need just a little bit more optimisation.
The transition into the fabs will probably be very, very fast. What’s needed is constant homogeneous electrical properties, which we can provide with MOCVD.
A 2-inch Ga2O3 wafer with an epitaxial layer from NextGO Epi on top.
RS: You have spun out of IKZ, an institute with expertise
in crystal growth. So why don’t you produce your own substrates?
A.F: First of all, as a founding team, our main expertise is epitaxy. None of us has ever worked on bulk growth.
There are IKZ experts producing the best and highest-quality material. However, in Germany, energy costs are not as low as in other countries, and it’s a high-energy process. You grow at 1,800 °C, and you have to keep that temperature for at least a week or two to grow one crystal.
The technique developed at IKZ is not the cheapest. You have a very bulky iridium crucible. With the iridium price fluctuating, when we put everything in the business model, we didn’t see that we would be compatible with other companies on the global market. So, we decided to stick to epitaxy.
RS: IP matters in our industry. NextGO Epi has two international patents. What do they cover?
T-S. C. One is a very unique technology to monitor the epitaxial growth of gallium oxide. The second is a process and technology patent for growing very thick and high-quality gallium oxide films, which are critical to power electronics applications.
RS: How do you characterise your material prior to shipment?
A.F: In the semiconductor world, reproducibility, reliability, and the specs are the keys to success. We have strict quality standards before we ship any product. We look at X-ray diffraction measurements, then perform Hall measurements to gauge the charge-carrier concentration, as well as the mobility of the material, because those two are, for most devices, the most important parameters.
Charge-carrier concentration gives you insight into how well you match the doping concentration to what the customer wants, and how clean your material is.
Mobility gives you direct feedback about the quality of the material. The higher the charge carrier mobility, the fewer defects and scattering centres are in the material.
We do some general screening, with AFM and microscopy, to see that the whole surface is clean and fine.
RS: Are all your customers capable of processing your material into devices? Or do some work with foundries? And do you think fabless firms could prosper in a gallium oxide power electronics industry?
A.F: Back in 2016, when we gave the first samples to make devices, they just used their gallium nitride recipes. The first device directly broke the critical breakdown field record of silicon carbide and gallium nitride. So, this material directly exceeded 3 megawatts-per-centimetre. That shows you how compatible it is with the gallium nitride technology. Therefore, I think fabless foundries will have an easy way to get into the market, because they can go directly to any foundry that can process gallium nitride and is willing to take gallium oxide.
RS: Gallium oxide has weaknesses, particularly p-type doping. Why do you think that this is not a show-stopper?
A.F: Most materials still use an n-type channel, because electrons have a higher mobility than holes. But p-type is really important for field management.
At least two solutions are very promising and have shown 8 megawatts-per-centimetre critical electrical breakdown fields in the device. Actually, the periphery of the device broke down, not the gallium oxide, so people still think they can push even further.
One solution is an ultra-high-κ dielectric. There, you have polarisation charge that helps you with field management. This is, of course, only possible for low-frequency devices. But when we think about high-voltage devices that run at 200 kHz, this is possible to implement.
The other one is integrating hetero-epitaxy or deposition. For example, p-type nickel oxide or copper oxide, using sputtering, which is a very normal process in any cleanroom. That worked like a charm.
RS: The thermal conductivity of gallium oxide is around just a tenth of that for silicon carbide. Is that an issue for this industry?
A.F: This topic has been part of research from early on. When the value is so low, people directly noticed and said: ‘We have to do something about it, otherwise this material will never succeed’.
The research community came up with promising solutions. Any device has to extract all the heat from the top, because heat will never make it to the bottom of the substrate of the device.
The most promising solution, lately developed by Martin Kuball at Bristol University, is another CVD process. You deposit polycrystalline diamond within your gate stack and your dielectric stack. Polycrystalline diamond has such good thermal conductivity that it extracts basically all the heat out of the structure.
You can get certain advances in packaging. We had, early on, collaborations with ZF in Germany. They showed it’s possible to package in a way that you can get the heat out.
RS: You are already shipping material to customers in various countries. How much of this business is associated with device development in academia? And how much is industrial development?
T-S. C. Most customers are from the R&D departments of their companies, or academia.
The NextGO Epi Founder team, from left to right: Andreas
Fiedler, Ta-Shun Chou, and Andreas Popp.
RS: You have just raised €2 million in seed investment. Tell me about your investors. Why are they backing you?
T-S. C. Our lead investors are Vireo Ventures, Ultratech Capital Partners and IBB Ventures, alongside angel investor Boris Habets. They are backing us because they believe in our next-generation semiconductor material for power electronics.
RS: What will that investment be used for?
T-S. C. For a startup, it’s all about growing, growing, growing. Our team now is three co-founders, all experts in gallium oxide. In the next couple months, we are having a new colleague join us, an engineer, to expand our production capacity and to allow us to attract more customers to gallium oxide. Also, we will have an operations manager join us to support us in the daily administrative stuff. We already have many student workers on site. In the future, when we attract more customers, we’ll definitely expand our team.
RS: What are your plans for the next 18 months?
T-S. C. We should establish our brand as the leading gallium oxide epiwafer provider. But more than that, we want to push gallium oxide technology further, providing high-quality epiwafers to companies around the world, to accelerate gallium oxide technology.
We need to see a gallium oxide product. The gallium oxide epiwafer is not a product. A module for the EV car, or the AI data centre, that is a real gallium oxide product. And we want to help everyone in the community make it happen.
