Songbin Gong wins DARPA Young Faculty Award

The unrelenting pace of Moore's Law drives electronics to push processing power onto ever-smaller chip sizes. However, the more performance per unit area that can be squeezed out of a chip, the more power the chip guzzles in the effort.

For comparison, it's simpler to think of developing electronics as runners preparing for a marathon. On the first day of training, a budding athlete doesn't have enough stamina to get even halfway done, and is so famished she scarfs down an entire pizza. Eventually, the runner improves her performance and starts regularly eating more to balance out the extra calories burned. 

Electronics are progressing in much the same way: as their performance with respect to size increases, they demand more energy. Unfortunately, two issues arise as electronics’ power consumption skyrockets. Devices tend to use too much power in a world that is trying to conserve energy and combat climate change, and mobile electronics may soon reach a point where the power they need outpaces the amount that onboard batteries can store for long periods of time.

ECE ILLINOIS Assistant Professor Songbin Gong sees “increased power consumption by electronics as an issue that must be met within the next 10 years. To solve it, we will completely overhaul modern electronic architecture.”

Gong won the DARPA Young Faculty Award for a concept that aims to develop an integrated circuit design that replaces the transistor with a completely new technology, an innovation on a level not seen since late ECE Professor John Bardeen developed the transistor at Bell Labs in 1947. If successful, Gong's technology will add yet another milestone to the legacy of groundbreaking innovations from ECE ILLINOIS.

“Our idea is to replace the transistors that are the basis of modern circuit design by developing nano-devices that leverage high-quality factor resonant excitation phenomena to express information,” Gong said. 

This award, presented yearly by the Defense Advanced Research Projects Agency (DARPA), is designed to recognize untenured college faculty members the agency believes will be leaders in their fields. It provides them with mentoring, funding, and DARPA contacts for the research they do for the agency. The award also gives young faculty members structure for possible future DARPA-focused research to pursue when they are established in their fields. 

DARPA provided award applicants with 17 topics, and faculty members were free to submit any proposal idea that fit into one or more of these areas. The organization provides $500,000 in funding for two years, with a possible extension of $1 million in total.

Gong’s proposal to DARPA involved an idea to significantly reduce the power consumption of modern computing systems. Traditional CMOS (complementary metal-oxide semiconductor) technology, which is the most common design used in integrated circuits and relies on millions of positive and negative-type transistors for its logic functions, inherently wastes much of the energy it’s provided with.

There exists no conventional way around the issue of power dissipation in traditional CMOS circuits, simply because the base design is inherently wasteful of energy. The only solution is to completely overhaul circuit design as the world knows it. Gong’s idea is to develop a system that dissipates very little energy as waste heat, and which can recover and reuse the most of the heat as additional energy. 

Gong’s idea to replace the transistor and overhaul CMOS-based computing systems is bold, he admits, but he is confident that the new technology will be both cost-competitive with current technologies, and also popular enough to eventually supplant CMOS-based designs. 

Gong will begin research on his concept in the fall with a graduate and a postdoctoral student. 

“I think it will be very exciting, but that’s not to say there are no challenges as right now, it’s just a concept,” Gong said. “We still need to do some serious work to even start demonstrating simple logic gates with the new technology. However, once the technology is fully developed and integrated into the batch manufacturing process, we expect our brand to be just as scalable, cheap, and high-performing as CMOS circuits but at much lower power consumption levels.”