The ultimate goal is providing next-generation silicon carbide (SiC) devices that boost the performance-cost ratio.
Power electronics remains based mostly on standard silicon devices. While three-level and other silicon circuit topologies are emerging to improve efficiency, new silicon carbide (SiC) designs are emerging to meet growing high-power requirements for electric vehicles.
In interviews, power devices managers at Mitsubishi Electric US highlighted the promise of silicon carbide when compared with standard silicon implementations.
They said efficiency improvements can be achieved with hybrid technologies that combine silicon with silicon carbide. For example, silicon-based insulated-gate bipolar transistors (IBGT) with silicon carbide Schottky-barrier diodes achieve efficiency improvements with relatively minor cost increases. For many applications, this represents a compromise between cost and performance.
Without changing topologies, Mitsubishi engineers asserted SiC is among the only ways to significantly increase efficiency.
Silicon carbide remains considerably more expensive than silicon. Hence, it’s important to identify applications where economics keep pace with energy savings or some other technical advantage to justify the expense.
Mitsubishi Electric has focused on SiC for high-power devices primarily because they are vertical components that operate at higher voltages. “Gallium nitride is a material that we have some experience within our RF group. And we think it certainly has very useful applications in lower power applications,” said Adam Falcsik, product manager for power devices at Mitsubishi Electric.
“So far, our power device development has focused on silicon carbide, primarily because it’s better suited for higher power applications. And, so, we have device modules [in production] rated up to 1200 A, and we have voltage ratings up to 3.3 kV,” Falcsik added.
SiC technology is viewed as unproven and therefore risky by power engineers who tend to be conservative. Many would prefer to wait for evidence of reliable performance before taking the plunge, thereby slowing SiC adoption.
Indeed, Mitsubishi engineers note customers remain in a “wait-and-see” mode.
“If early adopters are successful with this technology, delivering the desired benefits, there’s going to be significantly more adoption. And I think we’re gradually working our way through that phase,” said Mike Rogers, a power devices application engineer at Mitsubishi Electric.
Design changes are required to make the best use of silicon carbide, resulting in substantial reworking of PCBs. The resulting designs must be capable of handling much higher operating frequencies, the company adds.
EV, Storage Apps
Automotive applications stand to benefit significantly from silicon carbide technology, especially for electric vehicle drive trains along with battery recharging, either on-board or at charging stations. For EVs, “there is a strong desire to reduce the size and weight of electronics,” according to Tony Sibik, Mitsubishi’s power devices manager.
“Silicon carbide helps in that effort both by shrinking inverter size [and] by increasing efficiency, thus reducing the size of battery needed for a given range.”
Energy storage applications on the scale of electrical utilities is another potential driver of SiC adoption. The sector is benefitting from the shift to renewable sources like solar and wind to provide power when the sun doesn’t shine and the wind doesn’t blow.
Providing power during periods of peak demand requires sufficient capacity to store energy and, therefore, more converters and inverters. Silicon carbide is a promising candidate for those power conversion steps.
As more alternative energy sources come online, power flow requires special attention, including active filtering and harmonic correction. All require power semiconductors. Meanwhile, wide-bandgap SiC technology promises to boost storage of renewable energy.
One reason is that SiC delivers dielectric strength 10 times that of silicon, thereby offering a framework for building devices operating at higher voltages while meeting field requirements for remote charging infrastructure and smart grid applications. Moreover, higher switching frequency allows designers to reduce the physical size of magnets, inductors and other filter components, including transformers.
Mitsubishi Electric engineers note that silicon IGBTs, in general, have relatively slow switching, slowing further as the blocking voltage increases. IGBTs in the high voltage range, such as 3.3 kV, are quite slow and exhibit high switching losses, limiting them to low switching frequencies.
“Silicon carbide offers its advantage for 3.3-kV and, shortly, 6.5-kV devices,” said Eric Motto, Mitsubishi’s chief engineer for power devices. “More importantly, they can switch at considerably higher frequencies than a silicon device ever could.
“We’re seeing this today in… subway applications. We are mass-producing 3.3 kV silicon carbide devices for that application. They’re still pretty expensive devices, but the efficiency improvements they get not only in the inverter, but in other components of the powertrain make them applicable,” said Falsick.
Low harmonics due to higher switching frequency allows for significant improvements in motor efficiency, enabling wider adoption of SiC technology in high-voltage power applications. Mitsubishi Electric believes high-voltage DC transmission is pushing the limits of silicon devices, making SiC a more attractive option for those applications.
Higher device costs could therefore be offset by energy savings ranging as high as tens of thousands of watts.
SiC devices, especially at high voltage, provide faster and more efficient switching. Considering conduction losses, the best silicon IGBT is limited to about a 1.2-volt drop, even if operated well below its rated current. Silicon carbide exhibits almost no voltage drop at low currents, depending on how much chip area is used.
Mitsubishi Electric’s development roadmap implements optimization and new structures to improve SiC performance. “On the other hand, silicon IGBT technology doesn’t have much left to improve, where we’ve optimized that technology so much that it’s up against the physical limits of silicon,” Rogers said. “There are still some incremental improvements, especially in terms of optimization that is possible, but nothing as dramatic as we can achieve with silicon carbide.”
Mitsubishi Electric expects silicon carbide to remain more expensive than silicon for some time. Hence, early applications must justify cost via improved efficiency.
The strategy targets applications “that benefit the most, recognizing that any application that uses silicon IGBTs today could be more efficient using silicon carbide MOSFETs, said Sibik. “Att some point in the future, Si IGBTs will be completely obsolete—but how far into the future is still quite unclear.”
Schottky diodes also offer benefits. Mitsubishi Electric produces SiC Schottky diodes from 600 volts to 3.3 kV in high-volume applications like traction inverters requiring high current. DC-to-DC converter applications also require a diode, meaning SiC could provide power factor correction.
Long term, the goal is providing next-generation SiC devices that optimize the performance-cost ratio. Citing volume and competitiveness with IGBTs, Mitsubishi Electric acknowledges cost optimization will be critical for fine-tuning wafer process phases to support ever-growing production volumes.
Among the technical hurdles is the quality of the substrates made from silicon carbide wafers. Wafer defects continue to hinder yields. Those defects translate into higher SiC devices costs that ultimately hinder adoption.
This article was originally published on EE Times.
Maurizio Di Paolo Emilio holds a Ph.D. in Physics and is a telecommunication engineer and journalist. He has worked on various international projects in the field of gravitational wave research. He collaborates with research institutions to design data acquisition and control systems for space applications. He is the author of several books published by Springer, as well as numerous scientific and technical publications on electronics design.