Silicon-carbide semiconductor produced by Bosch at its fabrication site near Dresden, Germany. (Image: Bosch)

Future electric vehicles will be more efficient, more powerful, and will be able to hold more energy in their batteries than today’s EVs. Those big “mores” require countless small improvements beyond the headline component — batteries. One of the richest target areas is power-electronics technology and components used throughout the EV ecosystem. A new generation of power electronics will be found in tomorrow’s EVs, charging stations, and related infrastructure components.

Burak Ozpineci has a front-row seat to how better power electronics revolutionize the automotive industry. Now the Section Head of the Vehicle and Mobility Systems Research Section at Oak Ridge National Laboratory, Ozpineci has seen how technologies such as silicon insulated gate bipolar transistors (IGBT) were fundamental to both the GM EV1 and the Toyota Prius.

A chief power-electronics component, the motor inverter, is integrated with the traction motor and gearbox in Schaeffler’s “3-in-1” e-axle. (Image: Schaeffler)

“[Those vehicles] happened because of the silicon IGBT,” Ozpineci told SAE Media. “Then people used IGBTs because IGBTs became a household name for power electronics in many applications. Then, in the early 2000s, we started looking at silicon carbide, and in the early 2010s, everybody started looking into silicon carbide devices.”

Tesla famously used a silicon carbide (SiC) drivetrain inverter in the Model 3 when it came out in 2017, and Ozpineci said it’s expected in the industry that, after 2030, most EVs will use SiC devices. So-called wide-bandgap materials, specifically SiC and gallium nitride (GaN) offer marked efficiency improvements compared with silicon semiconductors.

Burak Ozpineci, Section Head of the Vehicle and Mobility Systems Research Section at Oak Ridge National Laboratory. (Image: ORNL)

“When we first started in 2001, there were just a few devices available,” Ozpineci said. “They were brand new, they were expensive. But with the introduction of the silicon-carbide MOSFET, maybe close to 10 years ago, people started looking at those devices more seriously.” The race to improve continues. Tesla announced in 2023 that it is developing a SiC inverter that will use 75 percent fewer silicon-carbide (SiC) MOSFETs.

SiC is Today’s State-of-the-Art

Vishnu Medisetty, Director of Power Electronics for Bosch in North America, confirms that SiC devices are the mainstay of the company’s power-electronics portfolio. Bosch began with SiC around 2001, he said in an interview with SAE Media, and the trajectory continues strongly upward.

“We are already on our second generation [of SiC],” Medisetty acknowledged. “We are working toward our third generation and there are generational improvements in the efficiency of the devices. So the device efficiency improves — but also, there’s a huge cost driver behind EV applications. What we look at is trying to reduce the cost of the power-electronic device, be it a charger converter or be it an inverter that is driving the motor. We try to reduce the size of the die, the chip, extract more current out of it, need less cooling for it.”

While Tesla hasn’t specified just how it will reduce SiC devices in future power-trains, the industry also is working on replacing SiC semiconductors with gallium nitride once it can solve at least one challenge: The current moves vertically in SiC devices but laterally along the plane in GaN devices.

In 2021, GaN Systems offered this comparison of GaN performance versus silicon and SiC semiconductors. (Image: GaN Systems)

“There are no gallium-nitride vertical devices available,” Ozpineci said. “The challenge with them is they’re usually low-voltage, low-current right now. In a low-power onboard charger, or if you’re looking at other DC/DC converters — again, low-voltage, low power — you can use GaN, which will beat SiC at the power level. But the vertical GaN devices, which actually will have better properties than SiC, are not ready yet — and they might not be ready for another 10 years.”

Bosch’s Medisetty agreed, saying that for now, gallium-nitride doesn’t have the performance needed by the transportation sector. “[GaN devices] are quite a bit lower-voltage, not readily suitable for automotive electromobility applications, but there are other gallium nitride approaches which are more similar to silicon-carbide devices,” he said. “These have higher voltages, are more robust and more suited for electromobility applications.

Fourth generation of Bosch’s EV motor inverter. (Image: Bosch)

“Bosch is working extensively on gallium nitride that is more suitable for electromobility applications,” he added. “We see a potential for it, but it’s still further out and there is lot more ground to be covered on silicon-carbide optimization. There’re generations before we get there.”

Sandia National Laboratories started a development project in 2019 for better next-generation vertical GaN devices for high-power-density electric drivetrains. Testing both GaN JBS diodes and GaN MOSFET devices, the teams have so far discovered that the main problem is the low maturity of the two devices. Engineers have reached current limitations for both the MOSFET devices (where the problem lies with the current output) and the JBS diodes (where the limitation is in the reverse leakage due to the etch-and-re-growth process).

Sandia said in a summer 2023 presentation that it will conduct further tests once either a multi-amp device can be made (for MOSFET) or the leakage problem is solved. These new, vertical GaN devices, like so much of the research that the DOE supports, are being developed to hit three key targets. The DOE’s power-density target for 2025 is 100 kW-per-liter for a 300,000-mile lifetime power-train that costs, at most, $2.7 per kW. Ozpineci said that regardless of the technology the VTO is helping with, money is always top of mind.

Vishnu Medisetty, Director of Power Electronics for Bosch in North America. (Image: Bosch)

“It’s always the cost,” he said. “OEMs are focused on three important things: cost, cost, cost.”

Ozpineci said he worked with a team in 2017 to develop a roadmap for 2025 targets, which included cutting costs by 50 percent compared to 2015. At the North American International Auto Show in Detroit at the beginning of the project, he noticed that many hybrid vehicles there had an extra box in the trunk, reducing cargo area.

“Talking to the OEMs, we heard about their interest in power destiny, too,” he said. “At that point, it almost sounded like power density was as important as cost. So, we focused on improving the power density by eight times by 2025 compared to 2015. But over the years, though, any time I bring up power density, they always say it’s the cost. Power density is really important, but cost is still the number one.”

VTO’s Support for R&D

The DOE’s Vehicle Technologies Office (VTO) has a mission to support R&D efforts that “lower the cost and improve the performance of power electronics in electric drive vehicles.” It funds and works on dozens of power electronics projects to do that. A glance at the VTO’s Annual Merit Review submissions shows the many ways companies and the government are working on these next-gen devices.

Some are working on integrating more components — heatsinks, power modules, control circuits — into the inverter, with the benefit of reducing the overall size so it won’t require as much space on an OEM’s skateboard EV chassis. Capacitors in an inverter could also be made smaller. Some projects use AI to redesign heat exchangers to potentially cool devices better in smaller packages.

Since 2020, American Axle & Manufacturing has been working with the VTO to eliminate the need for heavy rare earth materials in its motors. So far, this project has designed a highly integrated 650-volt inverter to eliminate phase leads and has developed a sintering method to attach discrete SiC devices to heat sinks for better thermal efficiency.

Another approach could come from new ways to cool power electronics using dielectric fluids. In 2018, the DOE started a $1.6-million project to improve thermal management in EV power electronics as part of the DOE’s Vehicle Technologies Program. The project’s objective is to develop thermal-management techniques to help reach that 100-kW/L power density target.

In mid-2023, the National Renewable Energy Laboratory said it had demonstrated that a new two-phase dielectric fluid concept, designed specifically for EV cooling, could perform better than more familiar water ethylene glycol (WEG) systems when used with a custom SiC module.

Bosch’s Medisetty added that development of higher-voltage EV platforms may help to literally take some heat off power-electronics cooling concerns. “Heat is a factor of the current that you’re pushing. At 800 volts, you actually have a lower current; if all things being equal, the power levels are the same, the current is lower on the 800-volt side,” he said.

“Typically, you go to 800-volt because you are already at the limit of the currents that you can push through, so you want to go to higher currents. What we look at is to try to optimize the inverter design such that you can shrink the size of the inverter, the cooling circuit – and take cost out. That’s what we are focused on,” added Medisetty.

To make a difference, research and eventual production-vehicle components have to find a home in vehicles people want to drive and own. After decades of watching power electronics improve, Oak Ridge’s Ozpineci said he knows who will be pushing hardest for change.

“There are some smaller and newer companies that are willing to adopt these technologies faster than traditional companies,” Ozpineci said. “New companies showing up in the EV space, they don’t have all these investments for the old technologies, and they said, ‘Okay, well, there’s this new technology that might be really beneficial. Why don’t we start from there?’”

This article was written by Sebastian Blanco, Editor-in-Chief of Automotive Engineering magazine, SAE International. For more information, visit here .