Hot research topic leads to best paper award

ECE Professor Umberto Ravaioli and PhD candidate Zlatan Aksamija have been awarded a First Place Outstanding Paper award by the IEEE Region 4 Conference Committee for their paper titled, "Joule Heating and Phonon Transport in Nanoscale Silicon MOSFETs." The award was presented at the 2007 IEEE Electro/Information Technology Conference (EIT), which was held in Chicago May 17 through 20.

"Our research deals primarily with a generation of a transport of heat in very small silicon devices. We hope to gain a better understanding about how to deal with dissipating excessive heat in something like a modern microprocessor," said Aksamija. "Current generation microprocessors already have a very strong limitation in the amount of heat they can dissipate and get rid of effectively. So we’re trying to improve upon that."

Ravaioli said thermal effects are perceived to be the main limitation to producing innovative processors, and should be taken into consideration when creating such devices. In fact, that is their ultimate goal.

"A lot of the design now is going toward reducing the heat rather than improving the performance [of devices]," said Ravaioli. "The new generation of devices will try to do the same tasks with much less energy … more efficiency in mind, rather than more power packed in. These things go hand in hand – you cannot just improve the brute force of the processor, you also need to think about your overall reliability for preventing thermal failure. It is definitely clear as you make smaller and smaller devices that temperatures will go beyond what makes the device work properly."

Presently, the predicament of heat generation is one of the biggest difficulties for companies, said Ravaioli. Heat may simply be stored and temperatures may increase up to the point where the device will not function according to specification, and may ultimately fail by melting. Ravaioli said he considers this an enormously important problem.

"If you understand how a single element generates heat and in time propagates then you can hope to make more complicated models," he said. "Potentially this kind of research is very important in guaranteeing improvements in computer technology keep going at the rate that we have today."

Aksamija and Ravaioli have been studying this specific type of problem for nearly four years, and said the  award is an evolution of their group’s research and physical simulation of nanoscale silicon devices.

The award selection was made by the committee after the conference upon review of the papers. All of the around 150 submitted papers in the technical program were peer reviewed with respect to their quality, originality, and relevance.

According to the conference Web site, "the Conference focused on basic/applied research results in the fields of electrical and computer engineering as they relate to Information Technology and its applications. The purpose of the conference is to provide a forum for researchers and industrial investigators to exchange ideas and discuss developments in this growing field." The 2007 EIT was the largest and best-attended conference since it began almost a decade ago.

Aksamija is a recipient of the prestigious Department of Energy Computational Science Graduate Fellowship, which sponsored the work along with the National Science Foundation. 

There was a slight delay in finding out about the award since Aksamija was away for a summer internship. He found the award plaque with his university mail when he returned.

Aksamija said he was surprised upon hearing their paper won first place. "I had received some positive feedback after the presentation, but I didn’t really get any indication that this was going to happen," he said.  

Ravaioli agreed. "I think [our first place award] indicates how much the interest toward thermal processes is now a prominent concern of companies to plan for the future generation of integrated circuits," Ravaioli said.