Semiconductor chips aren’t just the “brains” of our computers and smart phones. A different variety of chip, called power semiconductors, transform, route and manage the energy that powers our gadgets and are equally as important. Making them faster and more efficient is just as critical to enabling next generation technologies as improving processor speed.
That’s what power semiconductor experts Chirag Gupta and Shubhra Pasayat, electrical and computer engineering assistant professors at the University of Wisconsin-Madison, aim to do. With $1.5 million in support from the National Science Foundation ASCENT program and $3 million from the Advanced Research Projects Agency–Energy’s ULTRAFAST program, they are researching new materials, device designs and circuits that will take electric cars, the power grid and manufacturing to another level.
What, exactly, do power semiconductors do?
Gupta: I like to call microprocessors digital chips, which pack the largest possible amount of transistors, sometimes billions, onto the smallest node. In parallel, analog chips, or power semiconductors, perform different functions. Sometimes they only have one huge transistor. They help convert electricity from one form to another, like converting AC to DC, stepping voltage up or down, or routing or regulating power.
And these devices are actually everywhere. For example, data centers are one big application, because if you improve efficiency a tiny bit, it will have huge energy savings. In electric cars, the batteries are at a fixed voltage, so you need transistors to provide power to your headlamps, your steering, and your infotainment center.
But in principle, they are very similar to digital chips; nothing is terribly different about them and they are manufactured in the same foundries. It’s just that they need to manipulate more electrical power than a typical microprocessor.
What’s the maturity level of this technology?
Gupta: Until the late 1990s, silicon was the only semiconductor whose chips were commercialized for power electronics. But then new semiconductors—gallium nitride and silicon carbide—came onto market. These are called wide-bandgap semiconductors, and can convert energy five times more efficiently than silicon and can handle higher voltages. People have already commercialized those and they are used in fast electric vehicle chargers and phone chargers. They have reached the mass volume market.
But that’s just the tip of the iceberg. We will see even more gains moving forward as the remaining challenges get solved.
Pasyat: So while we are pushing the material limits of gallium nitride, our lab is also going into what’s called the ultrawide-bandgap regime. These materials are even more efficient.
They include aluminum gallium nitride and aluminum nitride. We are also looking at diamond. So these are all identified next-generation ultrawide-bandgap material systems. No one knows which one is going to win at this point. It’s quite exciting, because then there is a lot of work that needs to be done.
Can you explain your new projects?
Gupta: For the NSF ASCENT grant, which we received along with ECE Associate Professor Dan Ludois, we are working on a relatively novel class of transistors called bi-directional transistors. These are useful in the power grid and electric cars. The inspiration for the transistors came from Grainger Emeritus Professor Thomas Jahns and Professor Bulent Sarlioglu’s work at UW-Madison. In our project, we want to develop ultrawide-bandgap transistors using aluminum gallium nitride and gallium nitride. Shubhra will develop the materials, I’ll work on the transistors and Dan will integrate the fabricated transistors in a circuit.
The goal of the ULTRAFAST project is to use aluminum gallium nitride and aluminum nitride transistors in electric grids to give them super-fast response times. Because if the grid can respond quickly to faults, it would help a lot with the challenges of integrating renewable resources. We are using a novel strategy to make the transistors switch fast without producing electromagnetic interference, or noise, which can damage the grid.
So what material do you think will be the next big thing in power electronics?
Pasayat: Our training was in gallium nitride. We are nitride people. So aluminum nitride is the one we feel comfortable betting our resources on. We have a new reactor that is going to be configured for aluminum gallium nitride and aluminum nitride depositions. And we plan to dip our toes with other collaborators in government and at other universities to work on diamond and gallium oxide as well. So we will work as a team and try to develop all these materials systems and see which wins.
Gupta: At this point, it’s more faith than anything. All three are excellent choices and all three have their pros and cons. We just have to see which works out in the due course of time.
Top photo caption: Assistant Professors Chirag Gupta and Shubhra Pasayat examine a new ultrawide-bandgap semiconductor material that could lead to improvements in electric cars, the power grid, manufacturing and many other technologies. Photo by: Joel Hallberg.