Choosing SiC or GaN over Si Comes with Many Considerations

As silicon power IC (Si) technology begins to run into limitations, engineers are seeking out alternatives to help them build smaller, lighter, and more efficient products. Wide bandgap (WBG) materials such as gallium nitride (GaN) silicon carbide (SiC) are both viable options, so it isn’t always clear which of the three technologies is best for any given application.

Both GaN and SiC have seen increased interest and adoption in recent years as silicon power ICs have demonstrated performance limits, especially at higher voltages – there’s not much more that can be done with the technology to reduce size, weight, and power loss. GaN-based power ICs look to be the inevitable next step because they operate in the MHz range (Si technology has maxed out at about 200kHz), while SiC’s primary advantage over Si is its much better thermal conductivity.

Si, GaN and SiC all have their benefits and drawbacks, so in the absence of a clear winner, how do engineers pick which power technology when there’s an overlap zone between the three, especially when that zone could potentially shift down the road as all three are improved?

The choice between Si, GaN and SiC is ultimately application driven, but there are many pros and cons engineers must balance to evaluate which is the best option.

GaN and SiC poised to pick up where Si left off

Silicon power devices have a long history dating back to the 1950s when they displaced vacuum tubes. By the 1970s, a new class of power devices emerged thanks to the development of metal oxide semiconductor (MOS) technology for digital circuits, which allowed for the widespread use of metal oxide semiconductor field effect transistors (MOSFETs) in high frequency applications with relatively low operating voltages of less than 100 V.

By the 1980s, the insulated gate bipolar transistor (IGBT) emerged. Its primary and appealing characteristics include a simple interface, high-power density, and ruggedness, making it the preferred power technology for medium and high-power applications.

But Si IGBT and power Si devices are reaching limits inherent in silicon, most notably when it comes to high temperature operations – they’re not usable at temperatures beyond 150 °C. Use of Si power devices at higher temperatures leads to power losses due to the increase of leakage current in the off. This decreases reliability significantly.

GaN and SiC, as wide bandgap (WBG) materials, are superior to silicon because of their inherent high electron mobility and higher bandgap energy. Wide bandgap compound-based transistors also exhibit higher breakdown voltages and greater tolerance for high temperatures.

The ability to work at higher temperatures – above and beyond 200 °C – are why wide bandgap GaN devices are appealing for power device applications because it allows for simplified cooling systems in environments such as hybrid vehicles. Generally, simpler systems cost less because they require fewer parts and materials. One of the advantages of moving to GaN from Si IGBT is that engineers can leverage some of the same packaging technologies without having to make significant changes in modules.

SiC power devices share some characteristics with Si devices in that they have similar turn-on and turn-off voltage requirements. But like GaN, SiC has the ability to operate at higher temperatures than Si with better thermal management, while also offering high switching frequencies, lower switching losses, greater efficiency, and reduced size and weight.

While GaN and SiC are both contenders to replace Si technologies for power devices, they are quite different from each other, and each material has advantages versus the other.  Not only does SiC operate at higher voltages than Si, but also higher than GaN, which presently targets voltages between from 80V to 650V to deliver medium power.

While both technologies have lower switching losses than Si and offer very high efficiency at maximum power density, GaN and SiC differ at the transistor level. GaN employs a lateral transistor, while SiC uses vertical transistors that are more robust, as they also leverage gate-oxide reliability and offer excellent ease of use. Both GaN and SiC transistors operate at higher frequencies and switch faster than Si, and they’re both more efficient because they dissipate less power due to their lower on-resistance. This has led to their adoption in automotive applications, such as hybrid electric vehicles (HEVs), EVs, and EV charging infrastructure.

Click image to enlarge

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

Si, SiC, and GaN all have their use cases

While both GaN and SiC devices are gaining traction against incumbent technologies such as Si IGBTs, all three are viable options depending on the application.

Si IGBTs remain an option for a variety of power devices, such as pulse width modulated and servo three-phase drives requiring high dynamic range control and low noise – Si IGBT reduces the level of audible noise while improving dynamic performance and efficiency. Other common applications include uninterruptible power supplies (UPS) and any application with power circuits that require high-switch repetition rates.

Si IGBT improves upon predecessors such as power MOSFETs due its low on-state voltage and offering superior on-state current density – a smaller chip size is possible, and cost can be reduced. Si still demonstrates an ability to improve overall efficiencies in devices such as switch mode power supplies (SMPS). Si IGBTs are well suited for DC to AC motor drive applications because of their ability to handle high current, switching speed, and low cost.

Some of the other characteristics of Si IGBTs is their conductance losses, thermal impedance, and a high voltage rating with a low voltage drop, which make them suitable for high-power motor drive applications.

Where Si starts to become a less favourable option is when a design needs to be smaller and lighter – that’s when it makes sense to look at SiC or GaN. SiC allows for smaller motor drive systems because it can be embedded into motor assemblies – a motor controller and inverter can share the same housing.

In addition to power density and efficiency, Si has a tendency toward thermal runaway – the temperature of the device can rise uncontrollably, which leads to malfunctions and ultimately to failure. This is an especially important consideration for applications such as manufacturing or electric vehicles, where high current and high voltage conditions are common.

SiC is more resistant to thermal runaway since it’s thermally conductive, which means heat is more easily dissipated; the result is more stable operating temperatures. This thermal conductivity reduces the need for additional cooling systems – the overall system size can therefore be smaller and hence cheaper. It also makes SiC better suited to environments where the ambient temperature is generally warmer, such as in industrial and automotive use cases.

The higher switching frequencies of SiC ICs make them preferable over Si IGBTs for applications that require accurate motor control, such as automated manufacturing systems that employ servo motors for precision welding, tool arm control and object placement.

While SiC switches faster than Si, GaN switches faster than SiC; and although SiC operates at higher voltages than GaN, it also requires high gate drive voltage. Whether GaN or SiC is the best replacement for Si is heavily dependent on the application. GaN offers superior power density, especially where space is limited, such as data centers that require switched-mode power supplies that also benefit from GaN’s high switching frequencies and efficiency that can’t be matched by either SiC or Si IGBT. SiC is better able to handle high-temperature and high-voltage applications, including power string inverters.

The consideration that is keeping Si IGBTs in the game is they are still cheaper than SiC, which means they’re a better option for cost-sensitive applications if their limitations aren’t a factor. However, even though SiC is more expensive, the overall system could be potentially cheaper overall because of the lower cooler requirements and fewer components – a design and cost-study analysis will reveal whether it’s best to move on from Si.

Si, SiC and GaN all have their own requirements

Whether you ultimately select Si, SiC or GaN for your power application, each comes with its considerations.

Key Considerations for Si IGBT:

●      Si IGBT doesn’t contain a freewheeling or body diode by design, but it’s an essential part because it protects the switch by providing a freewheeling path to prevent reverse current

●      Care should be taken when either selecting a suitable diode or opting for a component with an integrated diode die

●      Auxiliary emitters are commonly used to improve switching performance of an IGBT as they reduce the influence of stray inductances within the gate circuit

●      Designs usually require more than one switch; two are typically necessary within a frequency converter

●      The supplying voltage taken from the grid has to be rectified, while the DC-voltage level also needs to be adapted or stabilized

●      An inverter is usually used to convert the DC-link voltage into the desired AC system,

●      A simple diode rectifier can be used if that application doesn’t require regenerative operation

Key considerations for SiC:

●      SiC’s Voltage Drain to Source (VDS) characteristics are linear and as a result lower conduction loss at any point lower than the full rated current; this is important to keep in mind for applications drive cycle is mostly below the full rated power, such as EV drivetrains

●      SiC MOSFETs typically have VDS breakdown voltages several hundred volts higher than Si IGBT’s 1.2kV

●      Reverse recovery (Qrr) is a critical consideration for symmetrical designs

●      A Kelvin source pin is essential for maintain the high switching frequency advantage of SiC devices because it mitigates inductance due to internal bond wires of the MOSFETs

●      A negative gate drive is necessary for ensuring a hard turnoff

●      Like Si devices, high-frequency EMI can be attenuated by using high-frequency material and a capacitor for EMI suppression

Key considerations for GaN:

●      Depending on their internal structure, GaN switching devices either come in enhancement mode (e-GaN) and depletion mode (d-GaN)

●      e-GaN switches operate like a normal MOSFET and has a simpler architecture and packaging without a body diode

●      A d-GaN switch is normally “on” nature and requires a negative supply

●      Designers must be mindful of parasitic elements associated with the package such as common source inductance and thermal management because they can potentially limit the device performance

●      GaN devices conduct in reverse direction even though they don’t have a reverse body diode because they are inherently bidirectional devices and will start conducting as soon as the reverse voltage across them exceeds the gate threshold voltage

●      GaN devices are not avalanche-capable by design so it’s critical to have a sufficient voltage rating with a rating of 600 V being generally adequate at bus voltages of up to 480 V for buck, boost and bridge topologies

As Si continues to hit its limits for some power applications, especially those where density and fast switching are required, GaN and SiC will become the preferred options. As long as there’s overlap among the three, engineers need to balance the strengths and weaknesses of each one for each power device application while keeping in mind the overall system cost.

Mouser Electronics