YORKTOWN HEIGHTS, N.Y. IBM researchers have developed a bulk process for producing nanotube transistors only 10 atoms wide, or 500 times smaller that current silicon transistors.
The nanotube transistors were enabled by an innovative patented process that IBM has dubbed "constructive destruction," which destroys defective nanotubes right on a wafer, rather than requiring that good ones be painstakingly constructed one-by-one, as is done today by other research labs.
"We believe IBM has now passed a major milestone on the road toward molecular-scale chips," said Thomas Theis, director of physical sciences at IBM's Thomas J. Watson Research Center here. "Our researcher's study to be published Friday (April 27) in Science magazine proves that we can grow carbon nanotube transistors in a manner similar to the way we grow silicon transistors."
Carbon nanotubes are but one of a plethora of molecular devices that researchers worldwide are developing to overcome the squeeze being put on silicon as design rules resist further shrinkage. Unlike silicon, carbon nanotubes can be grown with diameters of just 1.2 nanometers. Leading-edge silicon circuits are built with 130-nanometer (0.13-micron) linewidths. IBM's research team consisted of Phaedon Avouris, of the T.J. Watson Research laboratory, Philip G. Collins, a former IBMer now with Covalent Materials (Emeryville, Calif.), and Michael Arnold, an IBM intern from the University of Illinois.
"If we are ultimately successful, then carbon nanotubes will enable us to indefinitely maintain Moore's Law in terms of density, because there is very little doubt in my mind that these can be made smaller than any future silicon transistor," said Theis.
The particular problem facing bulk processing of carbon nanotubes was that all previous methods were spoiled because trace metals caused desirable semiconducting nanotubes to clump together with undesirable conductive nanotubes. The only successful methods, prior to IBM's constructive-destruction process, required tedious manipulations. IBM's patented technology, however, can distill out the non-metallic nanotubes in a manner similar to the way fuses can be blown on gate arrays to disable damaged silicon transistors.
"Three years ago the first carbon nanotube transistors were fabricated, but it was difficult to make attachments to the source and drain now those problems have been solved," Theis said.
The difference between the desirable and undesirable nanotubes is that straight nanotubes are metallic conductors, but the ones with an added twist become excellent semiconductors. These tubes are much narrower than is possible with silicon transistors. IBM said that today gates are several microns long and that short channel devices will require no new breakthrough, just a few more years of engineering effort.
To construct the nanotube transistors, IBM scientists first used vapor deposition to build mixed metallic and semiconducting nanotubes atop a silicon-oxide wafer, then attaching source and drain electrodes to metal pads over the nanotubes. In the next step, an electrode attached to the silicon wafer temporarily "switched-off" the semiconducting nanotubes to block the destructive current. Finally, the remaining unprotected conducting nanotubes were zapped with a high voltage, leaving the semiconducting arrays of nanotubes undamaged and ready to be formed into logic circuits like those found on gate arrays.
To make circuits, the electrical breakdown process was reused to remove individual carbon shells from semiconducting nanotubes with a precision down to a single device. By "blowing the fuses" one-by-one, the scientists hope to fabricate the semiconducting carbon nanotubes into circuits of field-effect transistors with any desirable band-gap.
Extending Moore's Law
"This a step on the road to extend Moore's Law, but now what we need to find out is what kind of performance we can expect from chips built with carbon nanotubes," Theis said. "Right now we don't know what their performance will be for sure."
With the new process, IBM will be able to determine within the next few years whether carbon nanotube-based transistors can achieve higher performance than silicon transistors.
"If we can prove that they can outperform any future silicon transistor, then IBM will bring the forces to bear to engineer a technology for mass producing them in complex circuits hopefully a chemical synthetic techniques that grows just the kind of nanotubes that we want, where we want them," said Theis.