The Tantalizing Potential for GaN in Electric Vehicle Power Electronics

­As electric vehicle (EV) power electronics undergoes a paradigm shift towards wide bandgap (WBG) semiconductors, it is clear that silicon carbide (SiC) is becoming the material of choice, while gallium nitride (GaN) is often shoe-boxed into telecommunications or optoelectronics applications. In part, this is because SiC has the greatest thermal conductivity (of Si and GaN), which naturally lends itself to the high temperature, power, and voltage operation typical of an EV. Yet, GaN can still achieve double the thermal conductivity of Si and is superior to SiC in almost every other metric, from electron mobility and efficiency to breakdown voltage.

The problem is GaN power devices today are operating some two orders of magnitude worse than their bulk material properties, reflecting the tantalizing potential. The stakes are also high as EV markets grow rapidly and OEMs are looking to improve drive cycle efficiencies using the WBG power electronics commercially available today. 

What's preventing GaN devices from tapping into this market? The crucial barrier is the material's production quality, which depends on the epitaxial substrate - GaN, SiC or Si. Since the primary source of material degradation are mismatches between the epitaxial growth and the substrate, the ideal case is homoepitaxy, or bulk GaN (GaN-on-GaN). Indeed, the knock-on effect on the performance when using silicon-based substrates is demonstrated by the blocking voltages achieved. Bulk GaN is 94kV, and SiC is 45kV, but GaN-on-Si, in volume production today, is around ~1kV because of mismatches, which is comparable to bulk silicon (Si). For a holistic view of the semiconductor supply chain, the new IDTechEx report "Semiconductors for Autonomous and Electric Vehicles 2023-2033" covers semiconductor trends and materials demand for electric vehicles, autonomous vehicles, battery management systems (BMS), radar, Lidar, infotainment and more. 

As is often the case with emerging technologies, the adoption of the best technology, in this case, GaN-on-GaN, is limited by the high cost. Bulk GaN is only available in small wafer sizes, contributing to a cost of around 1000 times greater than Gan-on-Si. The next best choice is Gan-on-SiC, which yields lower mismatches but again has a cost of around two orders of magnitude greater. Using GaN for high voltage applications, EV inverters, for example, therefore require either improving mismatches between GaN-on-Si or achieving low-cost production of bulk GaN. Until this is achieved, SiC will remain the dominant choice for high-voltage WBG applications.

However, opportunities are emerging and IDTechEx predicts that OBCs and converters will be the first market entry point, with timelines given in the report. This is because OBCs and converters operate at much lower powers and the efficiency advantage of WBG materials is a clear driver for faster AC charging or internal charging of the low-voltage battery (via the converter). 

Furthermore, there has been exciting progress for high voltage GaN in 2022 with new partnerships forming. VisIC Technologies, based out of Israel, is one to watch. The company develops automotive GaN power devices and partnered with Hofer Powertrain, which will use its 650V GaN chip in an 800V EV inverter design. This is one of the first examples of GaN technology being applied to a high-voltage inverter and represents a promising start. Given automotive adoption cycles are typically around four years, the door is opening for high voltage GaN adoption in EV markets, delivering a huge new growth opportunity for the industry.