Researchers at MIT have found a way to grow the carbon nanotubes that manufacturers need to build smaller, faster computer chips.
As chipmakers like Intel Corp. and Advanced Micro Devices inc. work endlessly to find ways to build smaller and smaller chips, they often run into a multitude of problems.
A key issue that must be resolved: dealing with the tiny copper wires that connect transistors in a processor. As the chips shrink in size, so do the wires, making it increasingly difficult for them to maintain the level of current needed to meet performance requirements.
That's where the nanotubes come into play.
"When we shrink chip features, the interconnections between the transistors get smaller - just like everything else," said Jim McGregor, an analyst at In-Stat. "And when they shrink, the performance isn't the same. Using carbon nanotubes is an interesting idea."
The question engineers have faced to date, he added, is "how do you do it?"
Gilbert Nessim, a researcher at MIT, said that chip manufacturers have consistently run into trouble by trying to build the nanotubes on a metal surface, which is needed to ensure an electrical contact.
Efforts to build carbon nanotubes on metal can face significant problems due to the heat required in the manufacturing process. The excess heat, for example, could cause the metals to form alloys that are not conducive to nanotube growth.
The MIT scientists used a combination of techniques to create a new process for using nanotubes. The techniques included vaporizing the metals tantalum and iron, which settle in layers on a silicon wafer. Then they placed the coated wafer at one end of a quartz tube, which was inserted into a furnace. The researchers also pump ethylene gas into the tube. The gas decomposes at high temperatures and the iron on the wafer catalyzed the formation of carbon nanotubes.
Nessim noted that the technique is based on processes already commonly used in the semiconductor industry. which should make them eaasier - and cheaper -- for manufacturers to adopt.
Intel, long looking to create the next generation of chips, is an underwriter of the MIT research. Nessim said the semiconductor industry has been interested in finding ways to replace copper wires with nanotubes, but has been slowed by the effort to find a reasonable way to deal with the metal issues.
"I hope this [research] may revise that enthusiasm to some point," he said. "At some point, they'll be stuck with copper that does not work and they'll have to find an alternative. We hope that our insights ... have eased a little bit of some fabrication issues and hope this will match the requirements that they have."
Intel, in a statement responding to questions from Computerworld, said they hope such research breakthroughs can help the company switch copper wires to nanotubes by some time after 2015.
McGregor suggested that a transition from copper to nanotubes could be some 15 years away. He noted that the shift would would entail some reworking of the manufacturing process.
"We won't go the way of carbon nanotubes until we've depleted current techniques," he said. "You don't go to something brand new and make a huge investment until you have to. We don't move at revolutionary steps. We move at evolutionary steps."
But McGregor did say that he thinks replacing the copper connectors with nanotubes would just be the beginning for the technology in chipmaking. Replacing the transistors with nanotubes would be the next step along that line.
"You won't replace the connections without looking to replace the transistors," he added. "The first step would be to replace the connections with nanotechnology. The second step is to replace the transistors."
The transistor, which acts as a switch, is the building block for the processor. It's even been called the most important invention in the 20th century.