NEW YORK, Dec. 10 (UPI) -- In order to keep computers advancing in power as they have for decades, a new U.S. research initiative partnering industry, academia and government has now launched to hunt in nanotechnology -- science and engineering on a molecular scale -- for a successor to today's dominant chipmaking method.
The stakes are extremely high in the race for leadership in this next era of electronics.
"It's not a God-given right for the United States to lead this," George Scalise, president of the Semiconductor Industry Association, told UPI's Nano World. "We have fallen off the trend line to making that happen, and we have to get back on it."
Scalise said because semiconductors provide the enabling technology for virtually all computation and communications systems, "leadership in semiconductor technology is essential to being competitive in the industries that drive the world's economy. Our ability to maintain a high standard of living, grow our economy, and assure our nation's security are all heavily dependent on being a leader in semiconductor technology."
The Nanoelectronics Research Initiative plans, over the next 15 years, to help create devices with features less than 10 nanometers -- or billionths of a meter -- in size, roughly 10 times smaller than in current state-of-the-art chips.
"We want a thousand flowers to bloom, to have lots of ideas out there," Pushkar Apte, SIA's vice president for technology strategy, told Nano World.
Today, the vast majority of chip manufacturing, in terms of dollar amounts, uses circuits based on CMOS -- short for complementary metal-oxide-semiconductor. The virtue of these circuits is they dissipate very little heat from the power they consume.
This, in turn, has for years allowed chipmakers to pack transistors closer and closer together without frying them, leading to the chip industry's ability to double transistor density every two years -- a trend dubbed Moore's law, after Intel cofounder Gordon Moore, that has remained true since 1975.
To keep Moore's law valid for greater and greater computing power, circuit features have grown ever smaller, with cutting-edge chips now boasting features only 90 nanometers in size, or less than a wavelength of visible light. The problem is CMOS is virtually impossible to push below 10 nanometers.
"When you get below 10 nanometers, you get to distances on the order of a few atoms, and hit fundamental physical limits no engineering can solve," Apte explained.
Circuits are based on electronic switches that flick on or off to represent bits of data as either ones or zeros.
"When you get as close as 10 nanometers, electrons can jump back and forth at random and you cannot maintain states of one or zero distinctly," he said.
Moreover, in packing tinier and tinier circuits together, heat simply cannot dissipate well enough, so as chips get smaller the danger of overheating grows.
"When the observation of Gordon Moore comes against the real laws in science, the law wins," Intel spokesman Howard High told Nano World.
Based on Moore's law, the 10-nanometer limit should arrive in roughly 15 years.
"The time for research to commercialization for most industries is about 15 years, from the first research paper to actual products. We think the time to start for a successor to CMOS is now," Apte said.
The mission will be to find the next switch to represent ones and zeros. The research initiative is planning to hunt for that magic switch in two phases. In the first, scientists will brainstorm for potential successors. After roughly three to five years, up to seven of the most promising ideas will be selected for more intense research and development in the second phase.
"The idea is to have a very simple prototype for demonstration at this proof-of-concept phase," Apte said.
He said one of the possibilities includes spintronics, which do not manipulate electron flow, as conventional electronics do, but rather affects the spin of electrons, thus circumventing the problem of power dissipation.
"DARPA (the Defense Advanced Research Projects Agency) has had a spintronics program for five years now," Apte noted.
Devices that manipulate molecules could be another possibility, he said. ZettaCore, headquartered in Denver, is developing memory chips that store data by adding or removing tiny amounts of electric charge from molecules.
Other technologies employ microbes to assemble nanowires and grow circuits. Cambrios, of Cambridge, Mass., is exploring such a concept, said Matthew Nordan, vice president of research at nanotechnology analyst firm Lux Research in New York City.
"Whatever company or companies have the winning technologies here will be immensely profitable to shareholders," Nordan told Wireless World, "but in public form certainly don't exist yet. It will be a daunting challenge."
The result promises to be nanoelectronic switches millions of times faster than existing switches.
"Of course, that doesn't mean entire systems will work that fast," Apte said, explaining that the switches are only parts of a system, "and a system is only as strong as its weakest link."
The initiative will be university-based and largely government-funded with industry direction and guidance, said John Kelly, a systems and technology group senior vice president with IBM.
"We're not talking about corporate welfare with federal subsidies to a company," said SIA spokesman John Greenagel. "We're talking about the government talking about funding basic research out of universities."
The initiative is partnering with the National Science Foundation. Apte said the aim is to work with other federal agencies with interests in nanotechnology, such as the National Institute of Standards and Technology.
"Our first order of business is to get consensus among all parties on defining and prioritizing the specific technical challenges on which to focus research," Scalise explained.
As time goes on, Kelly said, roughly three university-based national centers will emerge to carry out proof-of-concept work.
For phase one, the initiative will provide total seed funding of a few million dollars a year, while phase two will fund projects to a total of roughly $25 million annually. It is a drop in the bucket of the billions of dollars that likely will be required.
"On average, roughly 15 percent of revenues of the semiconductor industry are invested on R&D, which means today in the United States some $15 billion," Apte said. "In global R&D, if you assume the same percentage, it will be about $30 billion total."
Nano World is a weekly series by UPI examining the exploding field of nanotechnology, by Charles Choi, who covers research for UPI Science News. E-mail firstname.lastname@example.org