GaN Semiconductor Modules: Challenges and Future Opportunities

00:00:00: (ethereal music)

00:00:01: - "Sound on. Power on," your power electronics podcast,

00:00:06: powered by PCIM Europe.

00:00:09: - Hello, everybody, and welcome to the 11th episode

00:00:12: of "Sound on. Power on," your power electronics podcast,

00:00:16: powered by PCIM Europe.

00:00:18: My name is Marco Jung.

00:00:19: I'm a professor for eMobility and electrical infrastructure

00:00:23: at the Bonn-Rhein-Sieg University of Applied Science

00:00:25: at Sankt Augustin, as well as head

00:00:28: of Department Converter and Electrical Drives

00:00:30: at the Fraunhofer Institute for Energy Economics

00:00:33: and Energy System Technology at Kassel.

00:00:37: Both are located in Germany.

00:00:39: Additionally, I'm the chairman

00:00:40: of the IEEE joint IES/IAS/PELS German chapter,

00:00:46: IEEE PELS Region 8 vice chair, Germany,

00:00:49: and active member of several national

00:00:52: and international committees and communities.

00:00:55: And today, our technical theme

00:00:57: is GaN semiconductor models' challenges

00:01:00: and future opportunities, and this I will discuss

00:01:04: with Dominik Koch from the University of Stuttgart.

00:01:07: Hello, Dominik, how are you?

00:01:10: - Hi, thanks for welcoming me.

00:01:11: I'm fine, and thanks for asking.

00:01:13: How are you?

00:01:14: - Fine as well, so thank you.

00:01:16: You are a group leader and scientist

00:01:18: in view of obtaining a PhD degree.

00:01:21: I think that is a different way of looking at the things.

00:01:26: Please tell us a little bit of your work life,

00:01:28: and how it is possible to combine your group leadership

00:01:32: at the university and your academic goal.

00:01:35: - Yeah, that's, you can sum it up in two words.

00:01:38: It's quite a crazy work life,

00:01:40: positively formulated, is very diverse.

00:01:43: So there are a lot of different responsibilities,

00:01:45: especially in project acquisition and also in the,

00:01:49: let's say, the technical area

00:01:50: in supporting the different projects.

00:01:53: And this brings some advantages.

00:01:55: So I see topics, or I experience topics,

00:01:59: beside my actual PhD topics.

00:02:01: So, for example, we have projects which uses AI in,

00:02:05: for example, failure analysis or control,

00:02:07: and lifetime and reliability testing,

00:02:09: which were not original topics in my, let's say,

00:02:12: PhD thesis or in my PhD work.

00:02:16: And therefore, this is a very nice combination,

00:02:18: in my opinion, but it requires, of course,

00:02:20: a lot of good time management because you, yeah,

00:02:24: I mean a PhD is already more than a full-time job,

00:02:26: and then the group leader part is also quite challenging

00:02:30: regarding the time, but it works with, like I said,

00:02:34: good time management and also in the collaboration

00:02:37: with my colleagues, so the other PhDs,

00:02:39: and together with the professor

00:02:41: and the administrative project management,

00:02:44: this is the only way it's possible.

00:02:47: - Well, that sounds really good,

00:02:48: and I think so really hard stuff, in this case.

00:02:51: So can you tell us your theme of your PhD thesis?

00:02:56: - Yeah, so I'm currently in the writing verses,

00:02:59: and it will head in the direction

00:03:00: of intelligent GaN power modules.

00:03:02: So I did the research there

00:03:05: in low-voltage, high-current GaN modules

00:03:08: and how to integrate sensors, how to integrate a good way

00:03:12: of parallelization of GaN transistors

00:03:14: to reach higher output currents.

00:03:16: And, I mean, this intelligent word, intelligent branding,

00:03:19: is always a very right definition, but in our case,

00:03:23: it means that we want to bring sensor outputs

00:03:26: to a system level so that, for example,

00:03:28: a system-level designer can use the temperature

00:03:31: or current information from inside the module in the control

00:03:34: or over temperature protection.

00:03:36: This is somehow the goal of my PhD work.

00:03:39: - Well, that sounds really nice and fits really good

00:03:43: to our episode right now.

00:03:45: It's a different episode of "Sound on. Power on."

00:03:48: We discuss several topics related to power electronics

00:03:51: and future trends, especially, we addressed, in one episode,

00:03:55: wide-bandgap semiconductor at application level,

00:03:59: and now, we see it's a market,

00:04:01: arrived silicon carbide-based semiconductors, for example,

00:04:04: in PV inverters, and in the community,

00:04:07: everybody speaks from the alternative, GaN semiconductors.

00:04:11: And this is what we want to address today

00:04:14: in our technical theme.

00:04:17: So let's have a look into the future,

00:04:19: and let's assume that gallium nitride semiconductor models

00:04:23: are available on the market.

00:04:24: What are the technology advantages

00:04:26: compared to the silicon carbide,

00:04:28: and what are the disadvantages?

00:04:30: Can you tell us a little bit about this?

00:04:33: - Sure, I mean, it's quite clear

00:04:35: and also shown in many presentations,

00:04:37: but I think the main advantages of gallium nitride

00:04:40: in comparison to silicon carbide

00:04:42: is that you have lower switching and conduction losses

00:04:46: if you compare it to the same, let's say, die area.

00:04:49: So the well-known figure of merits,

00:04:51: and hard and soft switching,

00:04:53: and all different other figure of merits are better,

00:04:56: and therefore you can get potentially a higher efficiency.

00:05:00: And this better figure of merits also results

00:05:03: in a potential higher switching frequency,

00:05:05: and there, you just have to take a look

00:05:06: on what is already available on the market,

00:05:08: all these mobile and laptop chargers,

00:05:11: which are significantly smaller

00:05:12: than the silicon counterparts,

00:05:14: and the size reduction also comes from the smaller footprint

00:05:18: of the GaN dice.

00:05:19: So if you take a look on a GaN semiconductor

00:05:22: with the same RDS(ON) rating as the silicon

00:05:25: or silicon carbide transistor,

00:05:26: then it's significantly smaller.

00:05:29: There's no free lunch, so in my opinion,

00:05:31: the main disadvantages are that, again,

00:05:33: with the smaller footprint,

00:05:35: you have a higher thermal challenge.

00:05:38: So it's very difficult, or not that easy, to get the heat,

00:05:42: the lost power out of the transistor,

00:05:43: and therefore, the thermal design is much more crucial

00:05:46: than in the past with big transistors,

00:05:49: and especially, I think, for drive applications,

00:05:52: the avalanche reliability is one thing,

00:05:55: so since we have a lateral, typically, unipolar device

00:05:58: with gallium nitride, the avalanche is not that good,

00:06:01: as with SiC, and also, in the past, with discrete devices,

00:06:06: the gate voltage robustness, and therefore,

00:06:09: the gate handling, is a thing

00:06:11: where the designer have to take a look into,

00:06:13: which is not that easy as with silicon or SiC,

00:06:15: which have a higher voltage margin at the gate.

00:06:18: And, I think, at the moment, or up to now,

00:06:21: the voltage rating is limited, in comparison to SiC, again,

00:06:25: since it's a lateral device, in most cases,

00:06:28: you don't have the possibility

00:06:30: to scale the breakdown voltage to, yeah,

00:06:33: to any extent you want, and these are the disadvantages,

00:06:36: in my opinion, which are currently tried to be solved also

00:06:40: in research.

00:06:41: - Mm-hmm, I think, so that is the point why GaN models

00:06:45: are not available at the market nowadays,

00:06:48: or better in the mass production right now,

00:06:50: although we must solve a lot of problems,

00:06:53: and what are the challenges in making it possible

00:06:56: to design large power converters?

00:06:58: Do we need new packaging, new model concept?

00:07:03: - Yeah, so, I think, on model level,

00:07:06: you always have to have a trade-off

00:07:08: between the thermal and electrical performance.

00:07:10: So if you think of the classical modules

00:07:13: for silicon carbide, they typically have some kind of DBC

00:07:16: or ceramic base plate,

00:07:19: and there you have a very good thermal performance

00:07:21: with very low thermal impedances and resistances,

00:07:24: but if you take a look on the parasitics of the module,

00:07:28: this will hindering your design with gallium nitride,

00:07:32: and this is something you really need

00:07:33: for GaN to bring this full intrinsic performance

00:07:37: of the gallium nitride switch to the system level.

00:07:39: You need, on the one hand,

00:07:40: a very low inductive package design, but also,

00:07:42: like I mentioned before, with these thermal challenges

00:07:45: because of the small die area,

00:07:46: you need a very good thermally optimized design,

00:07:49: and this is, I think, the main challenge at the moment

00:07:52: because the embedding packages and all the stuff

00:07:55: which were shown in the past on the market,

00:07:57: they're very good in terms of the parasitic design,

00:08:00: the electrical parasitic design, but they have a lack

00:08:03: in thermal design, and in research, there are ideas

00:08:07: to bring this, let's say, both approaches together,

00:08:09: so, for example, with a more advanced embedding,

00:08:12: or so-called sandwich modules, where you try to cool

00:08:16: from both sides and have electrical connections

00:08:18: from both sides, but I think the main challenges there

00:08:21: is that it's still very expensive than the,

00:08:23: this classical approaches with molding or DBC,

00:08:26: and there is a lot of research going on,

00:08:29: and also some companies presented in the past months,

00:08:33: and also, but also years, some, let's say,

00:08:35: first demonstrators with these classical approaches,

00:08:38: but, like I said, there is no free lunch,

00:08:40: so we will always have to do a trade-off

00:08:42: between the thermal performance

00:08:44: and the electrical performance,

00:08:45: and I think this is, at the moment, one thing

00:08:49: where it still needs to do some research,

00:08:51: and also, some application in the markets are necessary

00:08:55: to really design such a model in the end.

00:08:58: - Mm-hmm, sounds interesting, and I think so we will see

00:09:02: in the, let's say, in the next years, some new solutions

00:09:06: on the market from industry as well from the universities

00:09:11: in this direction.

00:09:13: And let's go a little bit deeper to the chip level,

00:09:17: and, as I know, actual blocking voltage of devices

00:09:21: are 650 volts, more or less.

00:09:25: And we have a project at my university

00:09:28: and at Fraunhofer IEE, and combined with several partners,

00:09:33: and we are developing a PV inverter

00:09:35: with a power higher than 100 kilowatt amps.

00:09:39: And our solution

00:09:41: for this three-phase grid connected inverter

00:09:45: is to using the power electronics topology

00:09:48: to solve these problems,

00:09:50: but do you see higher blocking voltage in the near future,

00:09:54: and what are the challenges in this direction?

00:09:57: - Yes, definitely, so there is a lot of research going on

00:10:00: in this direction.

00:10:01: I think the, or in my opinion, the main direction there

00:10:04: is to go to vertical GaN.

00:10:06: So, as I said before, the current transistors you get

00:10:10: in the market are just lateral devices,

00:10:12: so you have the source/drain and gate contacts

00:10:14: all on the top side of the substrate,

00:10:17: and typically, a silicon substrate.

00:10:20: And there is a lot of research going on into direction

00:10:24: to make it like the classical SiC or silicon devices

00:10:27: with a vertical power flow, so that you have the source

00:10:31: and gate connection on the top side

00:10:32: and the drain connection on the bottom side.

00:10:35: So, I think, there, the voltage scaling can be done

00:10:37: by adjusting the drift layer thickness,

00:10:40: and therefore, this is, in my opinion,

00:10:43: will be one of the breakthroughs to go to higher voltages

00:10:47: for gallium nitride.

00:10:49: We are also looking, together with colleagues

00:10:51: from Fraunhofer, ERF, and Freiberg, and Bosch and Ambibox,

00:10:54: in the frame of a project

00:10:56: into lateral 1.2-kV GaN transistors.

00:11:00: And there, one issue is the substrate leakage current,

00:11:05: and therefore, also the breakthrough.

00:11:06: And there we try to improve this behavior

00:11:10: with different kind of substrates,

00:11:11: so in terms of the substrate technology,

00:11:13: there's also a lot of research

00:11:15: which can also maybe enable, in the future,

00:11:17: lateral devices with higher current rating.

00:11:22: I think the oldest development in terms of vertical devices

00:11:26: is mainly later, and also the high power region,

00:11:29: so, for example, as you mentioned,

00:11:30: this 100-kVA solar inverter, but also, I think,

00:11:34: in traction inverters, there is a lot of interest there

00:11:36: because if you can, for example, implement it also

00:11:39: in a silicon substrate, then the potential

00:11:41: to reduce the prices for traction inverters, in my opinion,

00:11:44: is very high because you don't need that high temperatures

00:11:47: as for silicon carbide.

00:11:49: But there are challenges, and, in my opinion,

00:11:52: one of the, also, big challenges

00:11:53: that you somehow lose the advantages of the HEMT structure,

00:11:57: so, I mean, in a lateral HEMT,

00:11:59: you have the two-dimensional electron gas

00:12:02: on top of the transistor, and this is somehow, will be lost

00:12:06: if you go to a vertical concept.

00:12:07: There are adoptions,

00:12:08: but the performance might not be the same

00:12:11: as with the lateral one.

00:12:12: And another challenge,

00:12:14: but this is also what I mentioned before

00:12:16: is currently under investigation, is that the substrates,

00:12:20: at least at the moment, tend to be quite expensive

00:12:22: because if often use GaN-on-GaN substrates,

00:12:25: which are very expensive in the fabrication

00:12:28: or different kind of substrates, and there, I think,

00:12:31: is the current challenges to provide very major substrates

00:12:35: and also cheap substrates to boost this development

00:12:38: in this direction.

00:12:40: - And, hopefully, we will find a solution in the next years

00:12:44: because I think the cost reduction

00:12:46: is a really hard stuff for the industry.

00:12:48: And semiconductor is one enabler then

00:12:51: for maybe new applications, as you said,

00:12:55: using gallium nitride.

00:12:57: So, and I learned in the past

00:13:00: that we will see, in the future,

00:13:02: additional bidirectional blocking semiconductors.

00:13:05: What can they enable?

00:13:07: - Yeah, I think they can enable a lot of new

00:13:11: or improved topologies like the current source inverter,

00:13:14: or there were some publications

00:13:16: where they showed direct AC/AC metrics converters,

00:13:20: where you can really reduce the chip count

00:13:23: or enable topologies which are not feasible

00:13:25: with discrete bidirectional transistors,

00:13:28: so, I mean you can create

00:13:30: this bidirectional blocking capability

00:13:32: with any kind of transistor if you connect them antiserial,

00:13:37: but if you want to have the same RDS(ON) again,

00:13:39: you would need another two additional devices in parallel.

00:13:43: And this is, of course, very costly, and nobody's doing it,

00:13:45: so the topologies, in my opinion, some of them

00:13:48: are already there, but they are not as effective

00:13:50: as with monolithic integrated

00:13:52: bidirectional blocking transistors,

00:13:54: and, I think, what they can really enable is,

00:13:59: and this is not only my opinion,

00:14:01: we still have a very high energy consumption

00:14:03: in the motor drive applications worldwide,

00:14:06: so in all kind of escalators, ACs and so on,

00:14:10: there are very inefficient motor drives in there,

00:14:13: and this consumes around 50% of the worldwide electricity.

00:14:17: And I think if we can, are able to have the very cheap

00:14:21: and also cost-effective and efficient, for example,

00:14:24: current source inverters in this medium to low power regime,

00:14:27: I think you can, there, enable a very good energy savings

00:14:32: in terms of the motor drives.

00:14:35: - Mm-hmm, and in this direction,

00:14:37: I will speak a little bit more with you

00:14:40: because every technology will certainly have its place

00:14:43: in the future.

00:14:45: Where do you see the future areas of applications

00:14:47: for EEF technology, yeah?

00:14:49: You mentioned something before, but for 650 volts,

00:14:53: for 1,200 volts, and for the bidirectional devices.

00:15:00: - Yep, I think they can cover a very, very broad range

00:15:03: of applications, so, I mean this is not only my opinion.

00:15:06: If you take a look on the reports

00:15:08: of the semiconductor companies,

00:15:10: or the market research institutes,

00:15:13: there are a lot of different applications.

00:15:15: I will extend this list to also the low-voltage transistors

00:15:19: because I'm somehow coming from the low-voltage GaN area,

00:15:22: and I think there, the main applications

00:15:25: will be in the 48-volt battery systems in cars,

00:15:30: all kind of auxiliary DC converters there,

00:15:33: to 12-volt for example, and also in the audio

00:15:36: and LIDAR area, there are a lot of applications,

00:15:39: and GaN is already also used there in a lot of use cases.

00:15:44: On the 650 volts side, I see the main applications

00:15:48: in the consumer market.

00:15:49: I mean, they are already there with these laptop chargers

00:15:52: and so on, and all kind of mobile

00:15:55: and data center applications, so the, I mean,

00:15:58: it's well-known that we have an increase

00:16:01: in all this artificial intelligence in the past,

00:16:03: and they consume a lot of computational power,

00:16:06: and therefore, a lot of energy

00:16:08: and different kind of server centers and so on,

00:16:10: and there, I think gallium nitride,

00:16:12: together also with the other semiconductors,

00:16:14: I mean, there will always be silicon

00:16:16: and there will always be silicon carbide,

00:16:18: but together with the other counterparts,

00:16:21: the GaN can really be a boost there.

00:16:25: And, in my opinion, also for smaller solar inverters,

00:16:28: I think that the use of GaN is still feasible.

00:16:31: Maybe not for the high-power ones, as you mentioned,

00:16:34: but in, let's say, in the small, some kilowatt range,

00:16:37: it makes sense.

00:16:39: If you take a look on, from the, let's say,

00:16:41: from the research science

00:16:43: or from the pre-development side, from companies,

00:16:45: I think that also in the onboard charge area,

00:16:48: there will be a use case for gallium nitride

00:16:51: mainly in the 400-volt batteries,

00:16:54: and maybe also in small traction inverters below,

00:16:57: let's say, 50 kilowatts for small vehicles,

00:17:00: scooters or similar, which also use a battery voltage

00:17:04: below 400 volt.

00:17:05: There will be use cases for gallium nitride

00:17:07: because there it's much cheaper than SiC

00:17:10: and much better performing than,

00:17:12: for example, in silicon IGBT.

00:17:14: For the 1,200-volt devices I see applications also

00:17:18: in the onboard charger.

00:17:19: Again, for 800-volt batteries,

00:17:22: solar will be a thing, and also industrial motor drive,

00:17:26: but there you maybe also have to differentiate

00:17:28: between a vertical and lateral GaN,

00:17:31: so lateral GaN, in my opinion,

00:17:33: will always be for a lower power level,

00:17:35: and as soon as we may have vertical GaN transistors,

00:17:39: there might also be a scaling to higher power levels,

00:17:43: in my opinion.

00:17:44: The bidirectional transistors, as I already mentioned,

00:17:47: in these current source inverters,

00:17:49: or in this direct AC to AC conversion.

00:17:52: But there, if you take a look on what the manufacturers

00:17:56: of the bidirectional transistors are saying,

00:17:58: it's currently some kind of a chicken/egg problem.

00:18:01: So they have the transistors, the use cases are there,

00:18:04: but no OEM is currently willing to start a design there

00:18:09: or, let's say, to produce it and,

00:18:12: or sell it in mass production

00:18:13: because there are still some maybe challenges to solve,

00:18:17: but someone maybe has to start first,

00:18:19: and this will also be used on this applications.

00:18:23: - So let's change our view of the things,

00:18:26: and let's have a look at the characterization

00:18:29: of the GaN devices and the modeling of the semiconductors.

00:18:34: What needs to be taken into account?

00:18:36: Please share your experience with us

00:18:38: because it's, I think so really important

00:18:40: for modeling, simulations, and so on.

00:18:44: - Yeah, this fits quite well to our work, what we are doing,

00:18:47: so, I mean, you need very good models

00:18:51: in all this kind of photo prototyping

00:18:53: and digital twin processes, design processes

00:18:56: which are discussed frequently because, I mean,

00:18:59: you can reduce the number of cycles

00:19:01: of a real hardware prototype if you have a good model,

00:19:04: but a good model needs good measurement data

00:19:06: because if you have bad measurement data,

00:19:08: your model can't be good, or it's modeling

00:19:10: not the transistor, but anything else,

00:19:13: and, in my opinion, what definitely has to be considered

00:19:16: that you question or somehow double check the results

00:19:19: of your model because, often,

00:19:21: models you get from manufacturers, they don't include,

00:19:25: for example, a breakdown or similar,

00:19:27: or the temperature is not modeled very well,

00:19:31: and therefore, this is, I mean this is always a trade-off

00:19:34: between the physical fidelity of a module

00:19:36: and the robustness, simulation time or convergence

00:19:39: of a model because you cannot have everything in one model,

00:19:43: and there, in my opinion, the designer

00:19:46: or also the characterization or modeling engineer

00:19:49: always has to do the trade-off between these parameters

00:19:54: and has to ask himself, "Okay, for what use case

00:19:58: do I need this model?

00:19:59: Do I want to make a system-level simulation,

00:20:02: or do I want to characterize device properties, overshoots,

00:20:08: or lost power in the transistor?"

00:20:09: And this is something which, in my opinion,

00:20:12: gets more and more important with GaN

00:20:14: because you need a very good understanding

00:20:17: and a very precise model to showcase this physical perform,

00:20:21: or the performance of gallium nitride,

00:20:23: and, I mean, the reason for this is maybe also well-known,

00:20:27: that devices can switch very fast, and therefore,

00:20:30: the bar parasitics, the measurement equipment,

00:20:33: are playing a crucial role, and then somehow are hindering

00:20:36: to extract the intrinsic design of the transistor,

00:20:39: and you need a very good bandwidth

00:20:41: and linearity measurement equipment,

00:20:44: and also the invasiveness, if you think of a current sensor,

00:20:48: plays a crucial role, and this is something

00:20:50: you really have to consider already

00:20:52: in the characterization process to somehow check,

00:20:55: okay, how am I affecting my transistor

00:20:58: with my measurement circuit, and there,

00:21:01: we are also looking into new measurement techniques,

00:21:03: or, for example, the last years,

00:21:05: the calorimetric measurements came up,

00:21:07: especially for the soft switching loss characterization,

00:21:10: but we also try to extend them

00:21:12: to hard switching evaluation because there,

00:21:14: you don't need to use the measurement equipment,

00:21:18: like current sensors,

00:21:19: and you can really have a very good design

00:21:21: in terms of parasitics and thermal design,

00:21:23: and this is, in my opinion,

00:21:25: something which also have to change in the future,

00:21:27: and also stuff like the normalization standardization

00:21:33: is playing a crucial role because the old norms

00:21:37: from silicon, they are not, maybe not the feasible anymore

00:21:41: to apply for at gallium nitride

00:21:43: because the limits doesn't make sense,

00:21:45: and the switching is much faster, so there will be a change

00:21:49: in all this characterization domain, in my opinion.

00:21:52: - Yeah, that's true.

00:21:53: That's my experience as well, but maybe,

00:21:57: I think it will be a combination

00:21:59: because with the calorimetric system, it's only possible

00:22:03: to, let's say, assume or measure the whole power losses,

00:22:07: and then you need really good models, yeah, to extract,

00:22:10: whether that's the switching or the dynamic losses,

00:22:13: and the static losses.

00:22:14: And so, I think so it will be a combination

00:22:17: of double pulse test as well as calorimetric.

00:22:19: What do you think about this idea?

00:22:22: - Totally agree, so again, I mean,

00:22:25: this is something you have to take into account

00:22:27: if you do a measurement, okay, what are they,

00:22:29: what do I want to extract from the measurement,

00:22:31: and, in my opinion,

00:22:33: so, we are still doing double pulse measurement

00:22:35: because they're easy, they, if you do them well,

00:22:39: if you know your measurement equipment,

00:22:41: you can still get very precise results,

00:22:43: but, again, you always have to take this into account,

00:22:46: and as you said, I think it will be a combination of both.

00:22:51: So you will have the hard switching double parts measurement

00:22:54: as, let's say, the basis, and depending,

00:22:56: if you want to have a hard switching topology,

00:22:59: you might not even need the calorimetric measurements

00:23:02: for the soft switching evaluation if you are saying,

00:23:04: "Okay, I will just have a soft switch converter,"

00:23:09: then you may skip the double parts measurement

00:23:12: because you just need the calorimetric

00:23:14: or the soft switching energy anyhow,

00:23:16: so, both methods will stay in the future,

00:23:19: and I think the future will show how to adapt them

00:23:23: or maybe combine them also in a nice manner

00:23:25: so that you have both in one setup, for example.

00:23:29: - Great, Dominik, in the last minutes,

00:23:31: we discussed the future of power electronics, but for me,

00:23:35: and I think so for the audience, it's really interesting,

00:23:38: what is about your future?

00:23:41: - Yeah, so the short-term goal is somehow to finish my PhD,

00:23:46: to finish my dissertation, because, I mean, it's now,

00:23:49: I'm working on it for five years now,

00:23:51: and now I should try to sum up everything,

00:23:53: but with all these very interesting topics

00:23:57: and all the projects we are doing, it's always,

00:24:01: you get distracted very easily,

00:24:02: and then somehow I have to try to focus on finishing my PhD,

00:24:06: and I don't made any decision.

00:24:09: Let's see what the futures bring.

00:24:11: I mean, I'm somehow open for everything,

00:24:13: and I think we have a lot of interesting topics also

00:24:15: in the future around SiC and GaN power electronics

00:24:18: and a lot of applications, in my opinion,

00:24:21: and, therefore, I'm very looking forward to the future

00:24:24: and the role of power electronics

00:24:26: and GaN and power electronics there.

00:24:29: - Great, thank you, Dominik.

00:24:31: - Thank you.

00:24:32: - To all the listeners, wherever you might be,

00:24:35: thank you very much for listening.

00:24:37: We hope you have enjoyed today's episode

00:24:40: and gained some valuable insights.

00:24:42: Make sure to subscribe to our podcast,

00:24:45: which is available on all major podcast platforms.

00:24:50: If you would like to share your feedback with us,

00:24:52: please do so via an email to pcim@mesago.com.

00:24:57: You can write to us also if there is a specific topic

00:25:01: you would like to be covered in the future,

00:25:03: or if you have a particular guest on your mind.

00:25:06: Until then, have a great time.

00:25:09: (ethereal music)

00:25:10: - We hope you enjoyed this edition

00:25:12: of "Sound on. Power on," powered by PCIM Europe.

00:25:15: Do subscribe and share.