Chip giant Intel is looking to developments in a tiny new form of science for the potential to lift future generations of its processors.
The chip maker is stepping up its research on carbon nanotubes, tiny tubes of rolled-up carbon hexagons, one field of the emerging science of nanotechnology.
Nanotechnology, which aims to use smallness as an advantage, is being explored by chip makers such as Intel Corp., IBM and others as a way to continue boosting the performance of processors and memory, leading to future advances in the chips that power computers used in critical roles by businesses.
Intel scientists are evaluating the use of carbon nanotubes in future transistors, the tiny on/off switches used to transmit electrical signals inside chips, said Rob Willoner, a technology analyst for Intel Research.
Nanotechnology, which by definition involves working with materials that are 100 nanometers or smaller—just a fraction of the width of a human hair—has become something of a buzzword.
Companies presenting at this weeks NanoBusiness Conference in New York are applying the concept to create tiny transmitters for communications gear, craft nanofibers that weave into sturdier clothing and even devise new treatments for diseases such as cancer. At the same time billions are flowing into nanotechnology research and development. The United States government, for one, will put nearly $4 billion into it in the next few years.
But, despite the funding, multitudes of startup companies injecting new ideas into the market, and the surrounding hype—and thats just for chips—some big names such as Intel are still cautioning that, although they do expect nanotechnology to lead to breakthroughs, it may be some time before they arrive.
Chip makers will continue to use silicon, which has long been the base material for chip manufacturing, for several more generations. Intels roadmap, for one, includes at least four more silicon-based chip manufacturing process generations, with the last one beginning in 2011.
But the ability to decrease the size of todays silicon-based transistors, which in turn allows chip makers to boost their chips performance by packing more transistors into each processor, will eventually hit a wall, leading chip makers to look elsewhere. (That date, which could change due to breakthroughs, is likely to be around 2020, according to the Semiconductor Industry Association.)
Thats the point at which carbon nanotubes, nanowires or other materials and manufacturing techniques made possible by nanotech research could come into play.
“For the foreseeable future, by which I mean well into the 2020s, things are going to be silicon-based. The industry has so much experience with silicon … and its such a wonderful material. There are a number of reasons” to stay with it, Willoner said.
“But youre going to see a lot of materials going on to that. Its a little bit like steel and concrete for buildings. Youre going to see that going on for a long time, but there are other materials that are always being added to a builders palate and thats what were seeing now” with silicon, he said.
Intel, of Santa Clara, Calif., IBM, headquartered in Armonk, N.Y., and others are still actively investigating that intersection. The companies are looking at using carbon nanotubes or other nanostructures to create smaller transistors.
Thus hybrid chips that combine a silicon base and carbon nanotube transistors are one possibility among many. Some chips might use traditional silicon-based transistors to process data, but use nanotech for memory.
One thing is clear, however. The fact that Intel, which relies on universities for its most advanced research and only later brings it in-house, has begun its own carbon nanotube research project shows that the company is getting more serious about that field.
The previously undisclosed Intel Research Strategic Research Project for carbon nanotubes, which consists of around 10 researchers, is charged with evaluating the nanotubes as a method of building transistors. They are exploring nanotubes ability to conduct heat in to see if the tubes can be used to improve chip cooling. The tubes could be added to the films used between chips and their sinks, to speed the removal of heat.
The project is still in its early stages and grappling with some fairly basic challenges, including ways of separating and assembling the tubes, a company spokesperson said.
Intel is also investigating nanowires, in part through an agreement with startup Nanosys Inc., as well as new types of silicon transistors. Those devices, dubbed tri-gate transistors, could be moved into production anywhere between 2007 and 2009, depending on the companys needs, Willoner said. Thus theyd hit the market well before any potential Intel nanotube transistor.
Tri-gate transistors have better control of their on/off state than standard transistors, meaning they dont leak as much electricity. Although there are techniques to combat it, leakage generally gets worse as transistors are made ever smaller. Theyre also easy to manufacture relative to other designs, he said. That could mean they stay around for quite some time.
IBM, which works with numerous other chip makers and also manufactures chips on a contract basis, is also investigating a broad range of nanotechnologies for processors and memory.
“We started working on carbon nanotube transistors in 1998. The first devices [transistors created from the tubes] were just curiosities,” said Tom Theis, director of physical sciences for IBM Research. However, he said, “The potential performance thats evidenced [by] that performance is greater than what we can see extracting from any silicon device.”
Despite being impressed with the nanotubes, IBM is also still evaluating its future transistor designs and has its own multigate silicon transistor, which it has dubbed a “double-gate transistor.”
“A lot of what weve learned will apply to any small device [or transistor], whether its based on nanotubes, nanowires” or something else, Theis said. “Were really not sure what the best [transistor] device will look like. Thats what were still exploring.”
Its safe to say chip makers will use whatever means necessary to continue providing generations of higher-performance processors and also higher-capacity, speedier nonvolatile memory devices to replace todays flash memory. Its through those developments that they aim to keep advancing business technology with higher-performing servers as well open up new possibilities, such as supercharged cellular phones with huge amounts of memory, Theis said.
“Were going to continue to make things smaller, faster and cheaper for some time to come,” he said.