Tiny tubes mean big chip advances
IBM researchers achieve a breakthrough the company says will help pave the way for the next era in the evolution of the microprocessor--beyond silicon.
The development in nanotechnology, the manipulation of molecular structures, will allow IBM to more easily create groups of transistors from tiny cylinders called carbon nanotubes.
IBM believes that nanotubes, which measure 5 atoms to 10 atoms wide and are 10,000 times narrower than a human hair, are the most promising replacement material for silicon in developing advanced chips in the coming decades.
IBM's achievement, to be detailed Friday in the journal Science, has allowed the computer giant to automatically produce pure semiconductor surfaces from nanotubes without the chip-frying metallic impurities that plagued earlier efforts.
Silicon has been the basis for manufacturing processors, memory and other chips for years. However, it is expected to reach its limits within 10 years.
That's because manufacturers must continue to shrink the size of the components printed on chips to cram more into a given space and thereby increase performance.
Silicon technology, which has been employed for years, is expected to last for at least another decade, thanks to breakthroughs in chipmaking techniques and materials. But chipmakers are preparing for the day when silicon will run its course.
Nanotechnology research, such as IBM's efforts with carbon nanotubes, is considered vital to the future of the semiconductor industry.
Nanotechnology materials are viewed by many researchers as the next big jump in the semiconductor's evolution because they offer smaller, more electrically conductive building blocks for processors. Nanotechnology would allow materials to be produced on a very small scale but in much higher volumes and at lower costs than traditional materials and manufacturing processes.
Sumio Iijima, a researcher with NEC, discovered nanotubes in 1991. Scientists later found ways to manufacture them in mass quantities.
But nanotechnology, like any new development, faces a number of obstacles before it will be ready for everyday use. The technology, for example, allows large amounts of materials to be produced inexpensively. But it also tends to produce more imperfections in a given area than silicon, said one researcher. This is just one issue that must be solved through research, or addressed with new ways of manufacturing.
But researchers at IBM and universities such as Harvard and MIT are taking a can-do approach.
IBM and other researchers had previously run into difficulties assembling transistors from nanotubes. The tiny cylinders, when synthesized, come in bundles consisting of both semiconducting and metallic carbon nanotubes.
The metallic parts, if left in the mix, would short-circuit a transistor. Without any other way of separating the metallic impurities, researchers were forced to assemble nanotube transistors by hand.
Silicon's successor
But while studying the effects of electrical currents on nanotubes, IBM discovered that with the right voltage, it could incinerate the metallic tubes, leaving only the semiconductors needed to build chip transistors. IBM believes that this work will lead to new research that will help establish nanotubes as the most worthy successor to silicon.
"We are already working on simple circuits," said Phaedon Avouris, manager of nanometer-scale science at IBM Research. These groups of nanotube transistors will ultimately become the ancestors of full-blown processors and memory chips based on nanotubes.
While solving the metallic issue is an important achievement, IBM and other researchers acknowledge that it's just one obstacle in a succession of barriers researchers must confront before nanotechnology chips are ready for the market.
"I would say a development effort will begin within three years," Avouris said. "A major issue is not just technological. It must be economically possible" as well, he said.
Nanotechnology researchers at Harvard applauded IBM's work, characterizing it as a major step on a long road.
"It will take several years and additional breakthroughs before nanochips become as prevalent as silicon chips are today," said Harvard Professor Charles Lieber.
A possible shortcut, Lieber said, is the development of hybrid silicon-nanotechnology chips. Hybrid chips could be ready in about five years, he said, while pure nanochips will likely come within a decade.
Hybrid chips would be extremely dense, allowing substantial gains in processor speeds or the amount of data a memory chip can hold.
One of the first hybrids will likely be a memory chip with fundamental elements, such as memory addressing, that use silicon. Meanwhile, the area of the chip where data is stored would be constructed using nanotechnology.
For hybrid processors, nanotechnology memory could be used to add a tiny, extremely dense memory cache to a server chip.
"We are already working on integration," Avouris said.
Nanotech-only chips would come after the hybrids were perfected.
"I think nanoscale is certainly the future of electronics," Lieber said. "But I do think there are a lot of significant problems to have addressed, with a significant impact on what's really going to (happen) in the future."