Table of Contents
Energy Demand is Increasing
“Save energy” is the mantra we’re all familiar with, but global demand isn’t going to fall anytime soon. According to the Industrial Energy Association, it will increase by about 50 percent until 2040 from 2018 levels. Optimistically, only two-thirds of that increase will be from renewable sources. A bit of mental gymnastics tells us that this means the actual amount from fossil fuels stays about the same. You might think that a future with more renewable energy would mean the efficiency of the conversion process would become less critical. Solar energy, for example, warms the environment whether or not you intercept and convert it to electricity, and eventually heat the load. Energy lost is still money spent unnecessarily, especially with the current higher cost from renewables, so with oil and gas still in the mix with solar, wind, and others for the foreseeable future, the efficiency of the power-conversion process from source to load remains a major issue.
Power Conversion: The Efficiency Challenge
Modern designs using resonant conversion techniques are now so effective that further improvements are down to basic component characteristics, particularly semiconductor switches. The ideal is that they are “off” or “on” in a “switched mode” design, in either case dissipating no power as long as “on” really is a short-circuit. The reality is that even a few milliohms of on-resistance can cause significant losses, and as the transistor slews between on and off states, it produces some transient dissipation. The level of transient dissipation can potentially reach the kilowatts range for a very short period. Keeping losses low therefore means getting that on-resistance lower and the device switching faster so the transient dissipation is for a shorter duration, and for a much lower average value. Traditional silicon-based switches such as IGBTs and MOSFETs are evolving, with incremental improvements, but new materials such as silicon carbide (SiC) and gallium nitride (GaN) are inherently better and are now the great hope for further improvements in efficiency.
SiC and GaN Wide Bandgap Devices Narrow Efficiency Gap
Compared with silicon (Si), SiC and GaN are different right down at the atomic level. The wide bandgap refers to the energy required to move electrons in the material from a “valence band” to a “conduction band” to be available for current flow. SiC and GaN have values about twice those of Si, and the effect on a device fabricated with the materials is dramatic. On-resistance is lower, switching is faster, operating temperature is higher, die areas are smaller and, particularly for SiC, thermal conductivity is far better than for Si or GaN. This means as a combination, less heat is dissipated, and what’s left is effectively led away, making for smaller and more efficient devices. There are knock-on benefits as well: better efficiency means less external cooling; switching faster allows other system components to shrink in size, reducing cost and product dimensions; driving the switches requires far less power than for competing Si devices; and SiC and GaN are inherently radiation hardened (rad-hard). This, along with their high-temperature operating capabilities, makes them suitable for aerospace applications. So, what’s not to like?
Wide Bandgap Semiconductor Uptake is Accelerating
Designers like SiC and GaN, but there are caveats: As they are new technologies, costs have inevitably been higher. These are reducing, and manufacturers are already claiming that overall lifetime cost is lower when system savings are factored in. Also, driving the devices is more critical than with Si and, in some cases, users are waiting for more reliability data before making the change from the more established Si technologies.
In the meantime, SiC and GaN device manufacturers are steadily moving along the evolutionary path, with wide bandgap technology recognized as having some way to go. On-resistances are reducing, voltage ratings increasing, novel packaging arrangements are being used to leverage the device performance to the maximum, and lab and field reliability data are accumulating. Even the sensitive gate-drive concerns are addressed with cascode arrangements of a SiC or GaN device co-packaged with a Si MOSFET for the best of all worlds.
SiC and GaN look to be the future for semiconductor switches, with efficiency gains approaching the theoretical limits set by practical interconnections. That is until the target moves again and the power engineer pulls another wide bandgap rabbit out of the hat.
