The State of Wide Bandgap

The Evolving Landscape of Wide Bandgap Semiconductor Technology

Image Source: krasyuk/Stock.adobe.com

By JJ Delisle for Mouser Electronics

Published June 8, 2023

Introduction

Virtually any technological system that can benefit from electrical control, communication, power, actuation, and sensing has been electrified and electrically connected. Beginning in the 1950s, the element at the heart of this technology has been silicon (Si), an incredibly versatile semiconductor with decades of development. However, Si is limited in specific areas of high power, high frequency, efficiency, radiation resistance, low noise, and optoelectronic capability. Third-generation semiconductors, especially wide bandgap (WBG) semiconductors, offer a performance advantage over Si that has justified the significant amount of time and effort required to develop economically viable semiconductor fabrication infrastructures and processes.

Many class IV, III-V, and II-VI compound semiconductor materials have wide bandgaps. These materials are often used for photonics, LEDs, and lasers, but only a few are suitable for broader semiconductor applications. The two leading WBG semiconductor technologies are silicon carbide (SiC) and gallium nitride (GaN). Though diamond (C) semiconductors have many attractive properties, the relative cost of producing diamond semiconductors has been a barrier to their wider use and applicability. SiC and GaN are increasingly used in high-power, high-frequency, high-efficiency, and high-radiation environments to enable performance levels that would otherwise be out of reach for other semiconductor technologies. Moreover, the maturity of SiC and GaN technologies has led to more significant economies of scale and larger wafer sizes, enabling cost reductions and extending the technology's suitability to many traditional Si power applications (Figures 1 and 2).

Figure 1: Comparing silicon (Si), silicon carbide (SiC), gallium nitride (GaN), silicon superjunction (Si SJ), and insulated gate bipolar transistor (IGBT)/gate turn-off (GTO) thyristor devices over power and frequency. (Source: Author)

Figure 2: The relative voltage capabilities of Si, GaN, and SiC devices as of late 2022. (Source: Author)

SiC in 2023

SiC is the most mature WBG semiconductor technology, supported by a few decades of development. SiC recently saw considerable inroads into high-voltage and high-power electric vehicle (EV) charging infrastructure. This will continue as EV and charging infrastructure manufacturers move toward 800V EV systems and beyond. SiC is also experiencing growing adoption in high-voltage direct current (HVDC) transmission and renewable energy applications where higher voltages enable smaller conductor sizes and lower conductor losses. As SiC is still roughly three to four times the cost of comparable Si technologies, it is viable only for applications where the much greater voltage capability of SiC justifies the cost. For example, the higher voltages enabled by SiC power electronics benefit photovoltaic (PV) solar and other renewable applications, allowing for the use of smaller and lighter passive components, such as inductors and transformers. This can result in net cost savings or a competitive advantage.

The major applications for SiC have been in high-voltage and high-temperature use cases such as power factor correction (PFC) circuits, AC-DC rectifiers, DC-AC inverters, battery chargers, and power electronics in data centers. With some predicting that 6-inch SiC wafer capacity—solely for the EV market—will grow from 125,000 in 2021 to over 4 million in 2030, this high year-over-year demand has led to substantial development in SiC manufacturing capability. The current target appears to be increasing production of 200mm SiC wafers, though future developments could result in larger wafer sizes that would further reduce the price of SiC devices and open the doors to other applications.

GaN in 2023

While SiC has gained traction primarily in 600V+ applications where traditional silicon has been less viable, GaN technologies have been actively replacing Si technologies in applications below 600V. GaN transistors have another unique advantage: Their electron mobility is many times that of Si and SiC. This allows GaN devices to operate at much higher frequencies. Hence, GaN provides WBG advantages at higher frequencies and power levels for switching devices and RF technologies. However, GaN semiconductors tend to have less than half of the thermal conductivity of SiC, which results in less power handling capability. This is one of the reasons designers prefer SiC devices for extreme voltages at high power levels (Figure 3).

Currently, there are indications that GaN power semiconductor devices are leading the market for applications requiring less than 400V. SiC power semiconductors appear to still lead for applications above the 800V threshold. GaN and SiC compete in applications between 400V and 800V, and this competition will likely grow in scope and voltage as higher-voltage GaN technology becomes more available. The competition between GaN and SiC at high voltages is certainly increasing with the introduction of 1200V GaN devices in late 2022. The power device market above 2000V is a relatively small market but will likely continue to slowly grow as larger industrial systems are increasingly electrified and more sustainable and renewable technologies are integrated into common grid power systems.

Figure 3: Distribution of GaN and SiC semiconductor devices across applications and voltages. (Source: Author)

Currently, SiC substrate and fabrication costs are higher than GaN devices, and final device costs of 5kW+ systems are comparable between GaN and SiC. However, SiC devices do have smaller die sizes than current GaN processes. Commercially available SiC devices already exist for 1200V and 1700V blocking voltage, with development ongoing for higher voltage devices; GaN transistors are rated 900V and, as of October 2022, 1200V. GaN devices developed for demonstration reach 1200V with comparable performance to SiC devices, but likely won't be available commercially until at least 2025. Some predictions indicate that the cost for 1200V GaN MOSFETs will be less than the expected 16 cents per ampere for SiC MOSFETs in 2025, as GaN’s simplicity and lower-cost substrate may lead to a cost advantage.

There have been ongoing efforts to develop larger GaN wafer sizes to further enhance the economies of scale for GaN. Other developments with GaN involve developing dual processes that support both GaN and Si on the same wafer. This would allow for high-density digital electronics to be developed on the same devices as power electronics and high-power RF electronics.

Other WBG Semiconductors

Many research groups are looking to develop other WBG semiconductor technologies, including gallium oxide (Ga₂O₃) and cubic boron nitride (BN), which are the most promising upcoming WBG semiconductor technologies (Table 1). These materials are used today in various applications, but they are still a long way from being used as commercially viable semiconductor devices. Many material properties of Ga₂O₃ have yet to be clarified, and there is an ongoing study of methods for developing Ga₂O₃ processes that may be viable for future fabrication.

The following list contains select common and future WBG semiconductor materials:

• Silicon carbide (SiC)
• Gallium nitride (GaN)
• Diamond (C)
• Boron nitride (BN)
• Zinc oxide (ZnO)
• Zinc selenide (ZnSe)
• Zinc sulfide (ZnS)
• Zinc telluride (ZnTe)
• Gallium (III) oxide (Ga₂O₃)/aluminum gallium (III) oxide ((Al₂Ga)₂O₃)
• Indium oxide (I₂O₃)

Table 1: Key Specifications of WBG Technologies (Source: Author)

BN is an auspicious WBG semiconductor material that is being studied mainly for optoelectronic and light-emitting applications. BN exhibits an indirect bandgap that allows for both p- and n-doping and has a high predicted breakdown field. It may also have a high saturated electron velocity and thermal conductivity and is one of the hardest known materials. Given predicted and simulated Baliga and Johnson figures of merit (FoM), BN could be ideal for power conversion and high-power and high-frequency devices. However, BN's doping states and practical characteristics are still unknown.

Conclusion

Both SiC and GaN are currently dominating the high-power and high-frequency markets, which will likely not change until other WBG semiconductor technologies are developed and become available commercially. This process usually takes nearly a decade; hence, we will continue to see SiC and GaN grow and compete at mid-power levels while SiC devices continue to lead GaN devices in higher voltage ratings for the next several years. Both technologies are significantly more expensive than Si, which may not change as SiC and GaN tend to compete at the fringes of what Si technology is capable of.