Wide-Bandgap Semiconductors for EVs – Panel


Manufacturers of electric vehicles and hybrid EVs are looking for efficient power-conversion solutions for several powertrain stages. Wide-bandgap semiconductors, such as silicon carbide (SiC) and gallium nitride (GaN), provide a performance edge over silicon in several respects: higher efficiency and switching frequency, along with the ability to withstand higher operating temperatures and voltages.

To enable electric vehicles to charge faster, automotive power electronics designers need GaN and SiC devices and a new powertrain architecture that can meet EVs’ efficiency and power density requirements. In order to obtain maximum driving range on a charge for the given battery capacity, the entire power conversion chain must achieve the maximum efficiency possible. Batteries must have a very high energy storage density. The autonomy of an electric car directly reflects the efficiency of its powertrain system.

The Roadmap to Next-Gen EV & AV conference offered vehicle designers insight into the building blocks for developing power efficient, advanced EVs with automated features. Speakers addressed the benefits of SiC and GaN during a panel discussion titled, “How Will Wide Bandgap Semiconductors Move EV Forward?”

The speakers were Mark Münzer, vice president, Innovation and Emerging Technologies team for the High Power business line at Infineon’s Automotive Division; Dilder Chowdhury, director of strategic marketing for Power GaN Technology at Nexperia; and Filippo Di Giovanni, wide-bandgap strategic marketing manager at STMicroelectronics.

Here are highlights from the panel discussion.

Zero Emissions

EE Times: As the automotive industry moves toward “zero emissions” transportation, manufacturers are rapidly ramping up their electrification programs. To meet customer expectations on performance these EVs require power electronic devices capable of efficient and effective operation at high temperatures. To meet those requirements, automakers and OEMs are turning to SiC and GaN technologies. Where can we find GaN and SiC in an EV? In which subsystems do we need a strong use of WBG materials, and which are the materials we cannot do without for the types of EVs we know today? Which changes is GaN/SiC offering to the automotive industry?

Mark Münzer: The market is booming; we are in a phase where new technologies are entering, helping to electrify vehicles on multiple levels. Looking at wide bandgap, naturally, we see that technology being able to increase the efficiency of the subsystems. The entry point is where the requirements meet the properties of the material, and both GaN and silicon carbide are extremely good when it comes to switching losses and on the partial load behavior. Therefore, naturally the first entry point here for wide-bandgap materials is the OBC, where, with the highest switching frequency, the [WBG materials’] switching behavior is definitely a plus. On the other hand, with the main inverter, it’s really about extending the range of the vehicle with the stored energy because in the end, this is what really makes the difference. There’s only limited space and a limited cost capability for the batter, so extending the range of the use of energy is a key driver for wide bandgap materials now and in future.

Dilder Chowdhury: “Wide-bandgap materials, in particular silicon carbide and gallium nitride, are coming into subsystems like onboard charger DC/DCs and eventually will end up in the traction inverters, where we are working on high-power configurations that will yield the most benefits. Here, the wide bandgap [materials] have very good switching performance, with very low switching losses, and very good high voltage performance. This is where it does much better against the traditional silicon superjunction or the IGBT solutions.

Filippo Di Giovanni: The strong use of wide-bandgap semiconductors depends on the targets that the designer has set for the vehicle. In other words, if the goal is extreme performance, as in the case of a sports car, or if we want to achieve the longest range per given battery pack, then a silicon solution for the inverter is a must; if we want to recharge the batteries faster, the OBC must be designed around silicon carbide and GaN. GaN will surely be the contender when it reaches full automotive-grade capability. So, for instance, if the electric vehicle is meant for a city car, then for limited range, maybe conventional IGBTs are most suitable.
Figure 1: EV block diagram (Source: STMicroelectronics)


Design challenges for the main subsystems

EE Times: In an electric vehicle, the traction inverter takes a high voltage from the battery and produces the power for the electric motor driving the car. The inverter controls the electric motor and captures the energy released through regenerative braking, giving it back to the battery. The DC/DC converter provides the 12-V power system bus, converting the voltage from a high-voltage battery. The efficiency of the inverter affects the longevity of the battery charge as it drives the motor. An HEV/EV includes several high-power devices. What are the design challenges for the main subsystems, such as the inverter, OBC, and motor? In terms of GaN and SiC, what are the parameters that designers need to consider when selecting the right device as well as the best topology for that type of application? What challenges do designers face when integrating the new power topology into their systems?

Chowdhury: With the on-board charger, for the PFC stage, we can take big advantage of the power gap devices, especially because there is no reverse recovery charge. That gives you hard switching and a totem-pole PFC configuration. There are advantages: First, you can reduce the component count, and at the same time, you can reduce your solution size. Then, in DC-to-DC converters, we have seen in our lab that gallium nitride devices outperform silicon carbide and, obviously, silicon. So, we can see the totem-pole topology for the PFC and on-board chargers, this AC-to-DC stage is a big advantage on the DC/DC by directionality, and even with the soft switching LLC, it gives improved efficiency and lower power losses. And there are some examples already out there. We are also working on a demo, and it’s  showing, actually, that you can achieve quite a significant improvement by increasing the frequency because of the lower switching and conduction losses.

Münzer: I think we have to see what key requirements of the applications are and this one is not as simple as simply saying that a main inverter needs to be efficient or an on-board charger needs to be small; we have to look really more into detail. If we need to have 800-volt fast charging, silicon carbide definitely is one way to go, or an IGBT might be a good solution for 1,200 volts. We strongly believe — and I strongly believe — in the coexistence of technology in an application. What you can see is coexistence of silicon carbide and silicon on the main inverter, especially if you have a drive train [that’s] four-wheel drive, [as I do]. I usually have one inverter that’s operating more than 90% of the time, but it is usually operating at very low power losses, so very much at partial load. And then the characteristics of silicon carbide are absolutely superior to the characteristics of an IGBT. So obviously, if I go for the main inverter on the rear axle, I would like to have silicon carbide. And as long as I can find silicon carbide from my battery, it’s very easy to make this decision.

Now, if I have a second axle, the front axle and usually the front four-wheel drive operate approximately 10% of the time, if at all. And in that situation, it’s usually operating at very high-power levels, and therefore it’s cheaper to go with an IGBT design. So, in the end — even in the same application— depending on my requirements, I may choose either technology and even combine them in the same vehicle.

Di Giovanni: Let me say that the challenge is all about efficiency. Now, better efficiency, as we all know, can be achieved by reducing conduction losses and switching losses. This means that for a silicon carbide MOSFET used in an inverter working at, let’s say, 15 kHz, on-resistance is the most important parameter, and not just at 25°C but also at higher temperatures, because this has also an impact on the cooling system. That system can be reduced in mass and volume, and let us not forget that in an inverter, the cooling system is a major pain in the neck. There is another big advantage because we can eliminate the bulky and heavy drive-to-machine cables. So let’s not forget that increasing the motor-phase number to reduce losses is much easier with an integrated system like this.

Regarding GaN HEMT, which is operated at far higher frequencies than silicon carbide, we know that GaN can easily work above 1 megahertz. It’s important to pay attention to the gauge charge and the capacitance, because this technology is primarily chosen for very high-frequency applications.

EE Times: To really take advantage of the new high-voltage WBG semiconductors’ benefits for EVs, packaging must meet many technical requirements to improve both electrical and thermal performance. What are the packaging considerations?

Chowdhury: Gallium nitride is very sensitive. It’s a very fast device. So, you need to have very low inductance in your packaging. What we are working on is all actually bonded technology, so we have no wire bonding associated with our packaging for automotive. It’s a very low-inductance package. And it also has a top-side and bottom-side cooling option, so that you can achieve very good thermal performance on the packaged devices.

Münzer: There are packaging challenges that come along with silicon carbide. First of all, for a given power rating, a silicon carbide device is approximately a quarter of the size of the equivalent silicon. This means that your contact area is only a quarter as large. So, your power density is going up to a level where potentially you start getting problems with the current capacity of your wire bonding. When we go to higher temperature, we will naturally get higher cycling and therefore more thermal mechanical stress. Even at the same temperature, we would get higher thermal mechanical stress with silicon carbide, as expansion coefficients for silicon carbide are more severe.

EE Times: A typical OBC architecture has a bidirectional front-end AC/DC stage followed by an isolated bidirectional DC/DC converter charging the high-voltage battery.  For the DC/DC stage of the OBC, LLC and LLC-derived bidirectional resonant converter topologies could be the preferred choice. What are the OBC design challenges with a look toward GaN and SiC?

Di Giovanni: A typical OBC architecture consists of a PFC stage followed by an isolated DC-to-DC converter. Now, this DC-to-DC converter needs to be bidirectional to implement, for instance, the vehicle-to-grid operation. Very often, LLC topology is used for higher efficiency with respect to a phase-shifted full bridge because the former enables us to realize zero voltage switching. Now, the problem with the LLC structure for bidirectional use is that when the converter is operated in the reverse power flow mode, the switching frequency is governed by the transformer winding capacitance and by the leakage inductance, which means that there is very little control — or no control at all — on the gain of the power stage and of the switching frequency.

So, one of the most used widespread topologies is the so-called CLLLC, with two capacitances and three inductances. In this topology, we can realize zero-voltage switching in the primary bridge and the zero-current switch in the secondary. The drawback of this topology is that the switching frequency needs to deviate from the series of resonant frequencies output voltage regulation. In order to overcome this issue, regulating the DC bus voltage in the PFC stage instead of frequency modulation is the most common approach. This viable DC link approach is very attractive because it enables the designers to reach a very high level of efficiency — around 98% — and the bus voltage is varied from 520 volts to 240 volts.

In general, gallium nitride and silicon carbide are both suitable for use in OBCs. But let us not forget that, again, silicon carbide exhibits a slightly larger energy gap and higher mobility; therefore, it can be switched at far higher frequencies. And gallium nitride can work without any problems at 1 MHz or even above. So, this is what we see today. Of course, silicon carbide is very attractive because of the short reverse recovery time of the body diode, even if VF in silicon carbide is a little bit higher than in silicon itself. But in the in the end, it’s a tradeoff, and all these disadvantages will disappear when GaN is used in the same topology at far higher frequencies.

Figure 2: comparison of technologies (Source: Infineon)

Figure 3: thermal simulation power GaN FET (Source: Nexperia)

Supply chain

EE Times: The whole market has been affected by Covid-19. In terms of manufacturing, what has been the impact on the WBG semiconductor market for EVs in particular? Have there been any changes on the supply chain? How are you organizing, and what short-term and long-term impacts do you foresee in the semiconductor industry?

Chowdhury: It’s a very challenging moment for all our factories, and our engineers are working in difficult times. This is also is also equally valid for our suppliers in the supply chain. But we are pleasantly surprised that, despite this difficult period, we are managing our supply chain very well. And it’s working without any reduction in our volume output. In many cases we have increased our productivity. So, the supply chain is obviously affected, but we have seen the suppliers adapting to the new reality. And obviously, some people are working in the factory and some are not, and they are adjusting to the new paradigm shift of working and trying to get the best out of it.

Münzer: I would say that our silicon carbide programs have not been impacted at all by Covid-19. Everything is going as planned, at the speed that we have expected prior to Covid. What changes the process in this specific area is the ambition of the automotive industry to ramp programs faster, to bring more vehicles to the market, which is a nice challenge to have.

Di Giovanni: Basically, the Covid 19 pandemic has disrupted the supply chain of silicon semiconductors for different reasons- As the vast majority of people have been forced to work from home, demand on PCs and tablets has skyrocketed. EV demand has not been affected at all by the restrictions put in place by different governments worldwide. ST has kept investing in increasing ST power silicon carbide capacity in terms of the front end, with the introduction of a new fab in Singapore in response to many ongoing projects with EV carmakers basically in all the continents. In addition, we secure wafer supply by signing strategic contracts with substrate suppliers.

The Wrap Up

In conclusion, can you give us an overview of how your company is pushing into the electric-vehicle market through GaN and SiC? Where are you currently seeing interesting applications that can drive the future of EVs? And where, in particular, do you think there will be significant changes in the near future to support your customers’ technical requirements for optimized performance and longer range with a smaller battery by using WBG semiconductors?

Münzer:  I think silicon carbide and gallium nitride still have a long road to perfection. We have just entered the market with our trench MOSFET and see a lot of potential here; we are already developing the second generation, and trench MOSFET, as everybody knows, is a way to go into the future. There is more for gallium nitride: It may play a great role in future hybrid systems.

Di Giovanni: ST has opted for planar technology and strongly believes it is still suitable even for the next generation of technologies. And this year, we will get to the third generation, which is an optimization of the second generation still in planar technology. We are also working with GaN. We acquired a majority stake in a French company Exagan, which by the way has realized a cascode design, and at the same time we are collaborating with a well-known leading foundry for GaN.

Although we recognize that gallium nitride is in an early stage of maturity compared with silicon carbide, we believe gallium nitride has huge potential. So, we are not only targeting the low power market; we are also targeting high power. I believe that our view of the world, from the lower power — say, from lower power right up to 130- kilowatt  — automotive market, GaN can participate very well. And obviously, we are bringing our third generation [to market], and for our fourth and fifth we are working to improve our specific RDS(on) and other parameters to enable more optimized performance from GaN.

Please visit the ebook for the complete article

Like it? Share with your friends!


What's Your Reaction?

hate hate
confused confused
fail fail
fun fun
geeky geeky
love love
lol lol
omg omg
win win


Your email address will not be published. Required fields are marked *