GaN Technology – Podcast – Power Electronics News

GaN improves the efficiency of power-conversion stages, making it a viable alternative to silicon in the manufacture of high-efficiency voltage converters. GaN has a number of advantages over silicon, including higher energy efficiency, smaller size, reduced weight, and cheaper total cost.

In this podcast with Andrea Bricconi, VP of Business Development at Cambridge GaN Devices, we will analyze the latest technologies of this wide bandgap ecosystem that will drive the next improvements.

Technical Articles

Guidelines for Thermal Management of GaN Transistors

GaN Technology: Challenges and Future Perspectives

10 Things to know about GaN

Transcription

Start

Welcome to Powerup, the podcast show hosted by Maurizio Di Paolo Emilio, that brings life to some of the stories on power electronics technologies and products featured on PowerElectronicsNews.com and for other AspenCore Media publications.

In this show, you’ll hear both engineers and executives discuss news, challenges and opportunities for power electronics and markets such as automotive, industrial and consumer. Here is your host, Editor-in-Chief of Power Electronics News and EEWeb.com, Maurizio Di Paolo Emilio.

-28:51

Maurizio Di Paolo Emilio

Hello everyone and welcome to this new episode of PowerUp. Today the topic is gallium nitride or GaN, and GaN transistors. GaN is a wide bandgap semiconductor, it is continuously growing in several power electronics applications. This is due to the exceptional properties of this material, which is better than silicon in terms of power density, resistance to high temperatures and operation at high switching frequencies. GaN greatly increases the efficiency of power conversion stages, serving as a valuable replacement for silicon in the production of high efficiency voltage converters. Compared to silicon GaN offers important improvements such as greater energy efficiency, smaller dimensions, lower weight, and lower overall cost.

Gallium Nitride semiconductor technology and applications are going to be a key highlight at this year’s Applied Power Electronics Conference (APEC) to be held on March 20th to 24th in Houston, Texas. Not only GaN, but also silicon carbide. In the previous episode, we talked with Peter Gammon about Silicon carbide technology. I will be at APEC and looking forward to meeting you there.

In this podcast with Andrea Bricconi, Vice President of Business Development at Cambridge GaN devices, we will analyze the latest technologies of this wide bandgap ecosystem that will drive the next improvements. The company is in the process of developing a range of energy efficient GaN-based power devices using its property, ICeGaN technology, to be deployed in key market segments such as consumer, switch mode power supply, lighting, data centers and automotive. Let’s talk with Andrea.

-26:57

Maurizio Di Paolo Emilio

Hi, Andrea! thanks a lot for coming on. How are you?

-26:53

Andrea Bricconi

Thank you, Maurizio. Very nice to talk to you. I’m very well on a sunny Munich finally.

-26:49

Maurizio Di Paolo Emilio

Good. So, today, the topic of this new episode of Power Up is GaN, gallium nitride, and GaN technology. But before starting, before going into details, tell us more about you. What’s your job? Please introduce yourself, your background.

-26:27

Andrea Bricconi

Thank you, Maurizio. My name is Andrea Bricconi. I have spent the last 25 years now in the semiconductor industry, in some of the top semiconductor manufacturers. In the last 10 plus years, I focus on gallium nitride technologies, applications, and marketing. And I have joined Cambridge GaN Devices or CGD for simplicity, a bit more than two years ago, and here I’m leading marketing and business development.

-25:57

Maurizio Di Paolo Emilio

Let’s start with the first question. So, why gallium nitride? as a material GaN appears to have significant inbuilt advantages over silicon for many applications, for many markets, which applications GaN is… already has significant market penetration and which applications are now emerging for GaN, gallium nitride?

-25:33

Andrea Bricconi

So, many questions! why GaN? well, I could ask why not? I mean, it is the most exciting material available these days. It has all to help the electronics making the next big step towards high efficiency and it is available and mature, finally, after many years of promises and setbacks. I will say that working in the field of gallium nitride is a real fun and you feel to be in a sort of middle of something special, which happens only once in a couple of decades. The market is booming, starting from low power AC DC applications, such as mobile chargers, adapters and in general all those switch mode power supplies for consumer applications. And, at the same time, we see that, after many years of preliminary activities, now there is good traction in server power supplies as well, driven by the needs to deliver the highest efficiency and comply with ever more stringent regulations. Besides, in the low voltage domain, let’s say below 300 volts, gallium nitride is helping reducing the form factor of DC DC converters in many applications such as robotics, drones, telecom, and by telecom I mean brick converters, just to be precise. So, wherever the need for smaller size and weight is really compelling, gallium nitride is entering the market and gradually replacing legacy silicon solutions. Of course, talking about emerging applications, GaN has a lot of traction also where silicon cannot cope with. Again, staying in the low voltage domain if [AB1] you think about wireless charging, that’s a nice example. Particularly with Class E, working at 6.78 MHz twice as much. The efficiency benefit of GaN is evident for[AB2]  those applications where you have to switch a big amount of current in a matter of few nanoseconds like in LIDARs for autonomous driving, simply there is not a valid alternative to GaN. So, I would say that, in general, all those applications where size and weight reduction is a target, because of specific market trends, like portability, or driven by international regulations, GaN is really becoming prominent. In terms of portability, if you think about an end user, I mean someone like you or me, we want to carry mobile chargers which are as light as possible, but as powerful as possible to be capable to charge many devices at once. And of course, enabling also fast charging. Of course, you can do, you can squeeze more power out of a charger with silicon, but it will be bulky, if you want to achieve very high-power density and very high-power capability, and a very small form factor, GaN is the way to go. On the other side of the spectrum on data centers, since we were talking about international regulations, and in general best use of energy, I mean, the electricity consumption of data centers targeted double by the end of the decade, achieving something like 800-terawatt power. We need to do the best use of that energy and GaN is definitely set to reduce the waste of that energy by improving the efficiency to the highest level.

-21:47

Maurizio Di Paolo Emilio 

So, currently, there are a couple of GaN device concepts. So, can you tell me which are the main and which is your direction into development from the design point of view? In an article, as you know, I have highlighted, ten points of GaN, ten things to know about GaN, is there a point where you are focusing with more attention?

-21:20

Andrea Bricconi

Oh, yes, I remember that article. It was something like “10 things to know about GaN. I remember, there’s one specific point I would like to address. But first, let me answer to the first part of your question: the available concepts and what we do about it. So, I would say that there are many concepts, much more than two, but somehow, we can talk about the extremes, the so called Cascode GaN and the so-called enhancement mode GaN. The cascode GaN was actually the first one which came to life thanks to my first company. I was there when International Rectifier, the pioneer in power GaN research, first started to develop GaN solutions based on cascode. And, by the way, almost every GaN manufacturer today has among its key managers, someone who shared that experience with me 20 years ago. We could say that IR legacy still lives in so many companies around the world. Now, by talking about the cascode itself, it is a two-chip solution at least two chips, where there is normally on GaN coupled with a low voltage MOSFET and the whole device operates through driving the MOSFET gate. Something which is well known, this concept brings ease of use. So, it is extremely easy to drive a low voltage MOSFET, so the cascode is universally recognized as an easy-to-use concept. Of course, it has some disadvantages. For instance, the lack of scalability. One thing is to make a cascode solution 600 volts or 650 volts. But if you want to do like a 100 volts device with that concept, the compromise introduced by the low voltage MOSFET is too big, you really lose the advantage of GaN. On the other side of the spectrum, we can talk about enhanced mode or E-Mode GaN: this is a wonderful solution, it is very elegant, a single chip normally off. But of course, many users would like to simply gain the performance advantage of GaN by replacing a silicon MOSFET without changing anything. Unfortunately, an enhanced mode GaN is very peculiar compared to a silicon MOSFET. It has an extremely low threshold voltage, just about one volt depending on the gate concept, but it’s basically between one and two volts, and the gate cannot see voltages above 6 or 6.5 volts. Therefore, the user shall either use a GaN dedicated gate driver or shall add a driving circuit which also clamps the gate voltage to prevent damages to the gate, or both. That is somehow the problem of this concept. And here, I’m coming to the second part of your question, your article. There was a point in your article where you mentioned that most of the E-Mode GaN solutions now require negative voltage to the gate for effective turnoff. And that is true because the threshold voltage is so low that you want to avoid any risks of PCB induced return on. So, not many gate drivers in the market have that negative voltage capability. So, you see that there are very good things about these concepts, but also some drawbacks.  Basically, CGD is coming with a technology that merges the ease of use of cascode with the simplicity of the E-Mode GaN avoiding all the drawbacks, including the needs for negative driving voltages, that is in a nutshell.

-17:41

Maurizio Di Paolo Emilio

So, your goal is to integrate the logic into an E-Mode GaN HEMT. So, it can be interfaced to drivers and controllers with minimal effort and also cost savings, so no need for extra components. So, your solution can be driven like a MOSFET. Can you tell me which are the reasons to integrate a logic instead of a GaN driver?

-17:09

Andrea Bricconi

Well, yes. 650 volts GaN technology by CGD is called ICeGaN. It’s a combination words. IC and eGaN, meaning E-mode GaN in one word, because in fact it is a system on chip where we have integrated logic into an E-mode power transistor. By the way, you can also read it like ICE GaN because it runs very cool. It does not incorporate a full gate driver, but just specific and key functions which enable easy interfacing with gate drivers. Years ago, when CGD was founded by Giorgio Longobardi and Florin Udrea, both from Cambridge University, after many years of consultancy with many companies involved on GaN, specifically on topics like GaN reliability and gate-related robustness, well they both mumbled a lot on the direction to go. At the end, what we wanted to develop was a concept fully scalable to high power and to low voltage. And we have decided that the best would have been to avoid integrating the fully gate driver but rather specific ICs to deliver key functions, to make GaN E-mode HEMTs as easy to use as a silicon MOSFET. And once you go into that direction, of course, the next step is to go further by integrating other functions like specific sense & control features for enhancing performance and reliability. And this is all at the core of our technology. Look, one of the things we have always heard about GaN HEMTs is that they are lateral, lateral transistors, and this is one of the key differences and advantages versus silicon and silicon carbide MOSFETs because this allows on-chip integration. CGD elaborated around this concept and focused all efforts to solve one of the major issues which prevented GaN really to find broad adoption in the last few years, to which I have alluded to a few minutes ago, that is the lack of ease of use. So CGD’s technology is an E-mode, that is a normally off single chip solution, which enables seamless coupling to standard gate drivers. Basically, you don’t need to add any RC network or clamping diodes whatsoever, you don’t need to deliver negative voltage to the gate. ICeGaN has a threshold voltage around 2.8 volts, it has an integrated Miller clamp which makes sure that under all circumstances the turnoff really occurs at zero volts, and that basically exploits the lateral nature of GaN HEMTs, to allow customers to drive GaN transistors like MOSFETs, up to 20 volts gate voltage. So, ease of use, we dare to say for the first time introduced on E-mode GaN, without any need of additional chips or any componentry, which by the way also saves costs and space on the PCB. So, I would say the concept is fully scalable. So, you can make a very high RDS (on) part for low power or very low RDS (on) part for a high power. It is scalable to low voltage by simply reshaping the design, but the concept stands, and this is how we basically think we have achieved our initial goal. The topic of introducing the gate driver, monolithically integrated into the power transistor, was considered, but I would say, this is an extremely good solution for low power and the market is telling that, today, we know. But it might become lossy at high power. I will say, due to the on-chip thermal coupling, the driver can suffer from extra losses due to the self-heating of the power transistor. And so, we thought it’s better to allow our customers to keep using the MOSFET drivers they have been using for years with silicon, to let keep the choice of the gate driver. Now that finally every gate driver can be coupled with an E-mode GaN: that is basically driving GaN like a MOSFET.

-12:37

Maurizio Di Paolo Emilio

So, a topic that I would like to comment now is in terms of thermal management, while wide bandgap semiconductors, gallium nitride, but not only, also silicon carbide solutions, promise higher operating temperatures and greater efficiencies. There are also as you know, thermal management issues that designers need to consider, when designing these devices into a system. So, what is your strategy regarding your technology, how do you see thermal management demands with increasing power density impacting the future development of process and packaging technologies?

-11:50

Andrea Bricconi

Well, yes, this is a good point. Of course, you know Maurizio, most of us involved in GaN often highlight one of the benefits of GaN is that one can achieve very high efficiency at much higher switching frequency than what is normal today with silicon-based architectures. In doing so, of course, we say that surface mountable packages shall be used in order to reduce inductances and support high frequency otherwise the package becomes the real bottleneck right. So, one cannot fully exploit the benefit of GaN. Unfortunately, there are not many SMD packages available for high power and people out there making power supplies are still relying very much on through whole packages like TO-220, TO-247. CGD is coming to the market with SMD packages like the DFN 8X8, DFN 5X6 for low to mid power applications. But it is very important also, especially for high power segment, in order to support high demand for efficiency in servers in data centers, so let’s say above one kilowatt, it is very important to develop thermally enhanced SMD solutions, and we are doing that with our assembly site. And they will have of course very low thermal resistance. We will share these in due time. In principle the topic is: with GaN we allow the user to shrink his application, but the heat is still there. It has to be dissipated somehow, are we heading into a dead end? So, I would say that first of all, we are on the safe side by using GaN because one should notice that GaN has by far the lowest switching losses. So, the output losses are significantly lower than any other technology. Gate charge is 10 times better than silicon, lower than silicon and silicon carbide. So, a partial answer is that with GaN we have to deal with much lower heat to dissipate. But of course, the topic of dissipating heat is important and I’m thinking that one of the directions we will see the market going in the next years will probably be with solutions like chip embedding for instance, with dramatically reduced thermal resistance via much more effective thermal management inside the PCB. But in general, I would say also with other elements coming from the GaN manufacturer that can help improving the cooling at PCB level. I said before that we are integrating sensing and protection features into the GaN HEMT. Think about the integrated current sense. Normally, in order to sense the current one needs to add external sense resistors, which of course prevents to connect the transistor to the ground plane. By bringing this function inside the HEMT, one can now connect the HEMT source to ground. And this way cooling can be much more effective because now you can design the cooling path according to your needs and focus on achieving either lower working temperature or vice versa, you can use a higher RDS (on) for the same thermals. So this is all that is included also into the ICeGaN under 50 volts technology. So, it’s a combination, the transistor technology can improve the cooling or deliver more degrees of freedom to the user. But of course, from a packing standpoint, there’s still a lot that can be done to help SMD technology becoming as performing as through hole.

-08:01

Maurizio Di Paolo Emilio

So, my last question is about a look into the future: How do you see GaN for the next years? What are the other wide bandgap materials to compete with GaN? So, I mentioned something about silicon carbide. So, in these days, we are talking a lot about electric vehicles too. So, where can GaN offer good value compared with other solutions and where do we expect to see the next wave of growth?

-7:29

Andrea Bricconi

Well, first of all, if you asked me that question, what is the future, a few years ago when the market was not there, I would say well, the market should start soon, because there are all the fundamentals. But now the market is really there. So GaN is rapidly gaining space, it is replacing silicon step by step gradually, but it’s booming at a CAGR of almost 100%. From consumer applications now ramping into servers, solar, and tomorrow of course, there’s no reason why you won’t be competing also for electric vehicles or hybrid electric vehicles. And silicon, of course, will keep dominating for many years. But the key turning point will be, of course, also connected to the product cost. When the product cost will be comparable, which is going to happen probably when the old industry, we will be working with 200-millimeter wafers, which is not yet the case, of course, at that point, we’ll see the fast replacing of silicon. Consider that, at system level GaN is already delivering much better cost than silicon, especially in high power application, but at product level, there is still a gap, which is going to be closed quite rapidly with in the low voltage domain, we’re almost there between GaN and silicon, already now. So, I think that GaN will keep increasing its pace, especially if we, as manufacturers, keep exploiting the peculiarities of GaN hat silicon cannot cope with, like bydirectionality, and further insist on integration. Of course, silicon carbide is there, and, by the way, it came earlier, and has proven a level of maturity, which GaN still has to demonstrate but silicon carbide, I mean, we all know that it’s rapidly taking over on IGBTs for traction inverters given the availability of 1200 volts devices, given its superior thermal performance, but there’s no reason why GaN cannot compete also, here. Perhaps you know that CGD is involved in a major European project called GaNext, with important IC manufacturers, magnetic manufacturers, the academia, and several end users for various different power levels. CGD is the sole source of GaN solutions. And the purpose of this project is to deliver prototypes and demonstrate power modules for low, mid and high-power applications. I really think that once GaN models will be available and reliable, well, entering the EV inverter market will basically be a matter of time and maturity of the supply chain. Coming to the last part of your question, you were asking where GaN can still deliver most of this potential? Well, if we look at the mega trends today, we probably have the answer. We’re talking about climate change, e-mobility, digital transformation, GaN demonstrates to be the technology which helps to deliver the highest efficiency among all of those. All of those who[AB3]  actually reached the manufacturing stage. I mean, I would bet… Personally, I would definitely bet on data centers to be the next big segment experiencing the advantage of GaN. I mean, it is so obvious, every 0.1% higher efficiency has a direct impact on the electricity bill. But there’s also a topic that’s really, there is a need, where why to waste less energy, which is becoming more and more precious. Moreover, server requires to dedicate more and more space to computing. So, saving the space for the power supply by making it smaller or squeeze more power out of the actual volume is a requirement, is a big requirement and GaN delivers exactly because of these needs. For instance, recently CGD launched a project funded by the UK Department of Energy called ICeDATA, aiming at designing and developing compelling solutions for data centers in terms of performance and reliability. So, I think that GaN will have a bright future. And I think also that CGD is ready for this exciting new era for power semiconductors.

-02:44

Maurizio Di Paolo Emilio  26:52

Thank you, Andrea. So, it has been a pleasure to have you in this podcast talking about GaN technology. Thank you.

-2:38

Andrea Bricconi

Thank you, Maurizio, and allow me to invite everybody who wants to talk to us, come visit us at the Applied Power Electronics Conference in Houston, Texas 20^th to 24^th of March. Thank you.

-2:25

Maurizio Di Paolo Emilio    

Thank you, Andrea. So, the market is booming, as Andrea said, starting from low power AC DC applications such as mobile chargers, adapters, and in general, all those switch mode power supplies for consumer applications. Now, as Andrea pointed out, there is good traction in server power supplies as well driven by the needs to deliver the highest efficiency and compliant with the more stringent regulations. There are many concepts of GaN devices. Cascode and enhancement mode, E-mode GaN. Cascode is a two-chip solution where there is normally on GaN coupled with a low voltage MOSFET and the whole device, as Andrea said, operates through driving the MOSFET gate. So it is extremely easy to drive a low voltage MOSFET and, as Andrea said, the cascode is universally recognised as an easy to use concept. E-mode is very peculiar compared to a silicon MOSFET: it has an extremely low threshold voltage. And as Andrea said, it works with a GaN dedicated gate driver or at a driving circuit. Cambridge GaN has integrated logic into any more power transistor. It doesn’t incorporate the full gate driver bat just specific and key functions which enable easy interfacing for gate drivers. According to Andrea, one of the directions we will see the market going into the next years will probably be with solutions like cheap and banding with dramatically reduced thermal resistance via much more effective thermal management inside the PCB. GaN is rapidly growing, Andrea bets on data centres to be the next big segment experiencing the advantage of GaN.

That brings us to the end of this episode. Stay tuned with more news and technical aspects about power That brings us to the end of this episode. Stay tuned with more news and technical aspects about power electronics. If you are listening to this on the podcast page at EEtimes.com, or PowerElectronicsNews.com, links to articles on topics we have discussed, are shown in this page. PowerUp is brought to you by AspenCore media, the host is Maurizio Di Paolo Emilio, and the producer is James Ede. Thank you everyone for listening. See you next episode. Stay tuned.