DAC 2013 -- Kaufman Winner Hu: FinFETs Will Serve Analog Design Very Well

Chenming Hu, TSMC Chair Professor at U.C. Berkeley, in accepting the 2013 EDAC Phil Kaufman award for outstanding contributions to the EDA industry, said the current issues about FinFET's applicability for analog design will fade quickly. 

"It's actually even better for analog circuits than for digital circuits in terms of benefits it's going to bring over the planar transistor," Hu said during his acceptance speech here Sunday night (June 2). 

"And it's going to be found to have better stability and reliability. That's another thing people do not realize. But if they think through the electric field in the transistors compared to planar transistors, (they're) going to see it's going to take between a half a volt to 1V of overdrive. And that's a world of difference." 

Hu, who climbed Mt. Everest when he turned 50, has spent his career "attacking the barriers to Moore's Law decade after decade," according to Klaus Schuegraf, group vice president of EUV Product Development, Cymer Inc., a former student who honored Hu at the ceremony here. 

Schuegraf pointed out that in the 1980s, there was the 1-micron scaling barrier. 

"Could devices scale beyond 1 micron?" Schuegraf said. "Through a combination of understanding the physics and coming up with the universal model with a set of equations….there was a solution."

The 1990s brought concerns about voltage scaling. 

"The question there was could dielectrics withstand large amounts of tunneling current without causing reliability impact to devices?"

Research determined that the problem could be overcome, he said. 

"The conclusions are all the same: Keep challenging the barriers and keep Moore's Law, which is driving incredible productivity, driving the EDA industry, driving the entire semi industry moving forward," Schuegraf said. "It's astounding." 

Hu became the 19th recipient of the Kaufman award not only for his work in FinFETs, introduced to the industry in 1999, but for his development of the BSIM model, the industry-standard transistor model for circuit simulation and for his impact as a longtime educator. Hu, who studied as a graduate student at Berkeley under Andy Grove, started teaching in the early 1970s as an assistant professor at MIT and has spent the past 37 years at Berkeley. He has also served as chief technologist for TSMC.

Hu, who said frequently the key in academia is finding smart students and just "letting them do their thing," noted that his students are in the process of modeling FinFETs in arbitrary shapes, in addition to the rectangular geometry that FinFETs are currently constrained to. 

Injecting some humor into his presentation, Hu said the shape might event look like this:
 

"Finfet enables all the complex designs in the very small geometries," said Cadence CEO Lip-Bu Tan. (Cadence acquired a Hu-founded company, Celestry, in 2003). Thanks to "the founder of FinFET," Tan said, "we're all going to enjoy (this work) profoundly."

Hu himself noted: "Finally, I think it can be scaled to the end of lithography, wherever that brings us to."

  

Related posts: