Shaping Up Carbon Nanotubes

A new method boosts the density of carbon nanotube bundles by five to 25 times, creating efficient conductors that could one day replace copper as the primary interconnects used on computer chips and even hasten the transition to next-generation 3-D stacked chips.

James Jian-Qiang Lu, associate professor of physics and electrical engineering at Rensselaer Polytechnic Institute (RPI), together with his research associate Zhengchun Liu, decided to investigate how to “densify” carbon nanotube bundles after they are already grown. The team discovered that by immersing vertically grown carbon nanotube bundles into a liquid organic solvent and allowing them to dry, the nanotubes pull close together into a dense bundle.

A carbon nanotube bundle before (left) and after (right) densification. (Photo: Rensselaer/Liu)

Lu attributes the densification process to capillary coalescence, which is the same physical principle that allows moisture to move up a piece of tissue paper that is dipped into water. He detailed the results of the post-growth densification project at the Institute of Electrical and Electronics Engineers’ International Interconnect Technology Conference (IITC), held last week in Burlingame, Calif.

Theoretical studies show that carbon nanotubes, if packed closely enough together, should be able to outperform copper as an electrical conductor. But because of the way carbon nanotubes are grown -- in sparse nanoscale “forests” where carbon molecules compete for growth-inducing catalysts -- scientists have been unable to successfully grow tightly packed bundles.

The new process boosts the density of these carbon nanotube bundles by five to 25 times. The higher the density, the better they can conduct electricity, Lu said. Several factors, including nanotube height, diameter and spacing, affect the resulting density, he added. How the nanotubes are grown is also an important factor that impacts the resulting shape of the densified bundles.

Images of the experiment are more striking than any before-and-after photos promoting fad diets. In one instance, Liu started with a carbon nanotube bundle 500 micrometers in diameter, shaped somewhat like a marshmallow, and dipped it into a bath of isopropyl alcohol. As the alcohol dried and evaporated, capillary forces drew the nanotubes closer together. Van Der Waals forces, the same molecular bonds that boost the adhesion of millions of setae on gecko toes and help the lizard defy gravity, ensure the nanotubes retain their tightly packed form.

The resulting bundle shrunk to a diameter of 100 micrometers, with a 25-fold increase in density. Instead of a marshmallow, they said, it looked more like a carpenter’s nail.

Side view of a carbon nanotube bundle end before (left) and after (right) densification. (Photo: Rensselaer/Liu)

“It’s a significant and critical step toward the realization of carbon nanotube interconnects with better performance than copper,” Lu said of the findings. “But there’s still a lot of work to do before this technology can be integrated into industrial applications.”

Despite his initial successes, Lu said the density results obtained are not ideal and that carbon nanotubes would have to be further compacted before they can outperform copper as a conductor. A cose-up photo, taken with a scanning electron microscope, reveals there are still large empty spaces between densified nanotubes.

A close-up of densified carbon nanotubes. Empty space between the nanotubes suggests the process to densify nanotube bundles could be further optimized. (Photo: Rensselaer/Liu)

The research team is exploring various methods to achieve ever-higher density and higher quality of carbon nanotube bundles, he said. integrated into industrial applications.”

Lu is confident that these densified carbon nanotubes, with their high conductivity, ability to carry high current density, and resistance to electromigration, will be key to developing 3-D computer chips. Chips used today can only shrink so much smaller, as their flat surfaces must have enough room to accommodate scores of different components. But the semiconductor industry and academia are looking at ways to layer chip components into vertical stacks, which could dramatically shrink the size of a chip. <br.densified>

Nanotube height, diameter and spacing affect the densification process. If the nanotube bundles are too short, like those on the left, there is no densification at all. If the bundles are too tall, like those on the right, bundles are not rigid enough and tend to stick with one another after densification. The middle region, where bundles are between 30 to 65 micrometers tall, demonstrates good densification. (Photo: Rensselaer/Liu)

“Carbon nanotubes are one of the most promising materials for interconnects in 3-D integration,” he said. Other potential applications of the densified nanotubes are high-surface-area electrodes for supercapacitors, fuel cell electrodes for hydrogen storage, heat dissipation materials for thermal conductors, and other situations that require high electrical, thermal or mechanical performance.

Lu’s group collaborates with a research group led by Pulickel Ajayan, the Henry Burlage a professor of materials science and engineering at RPI, who provided Lu with the grown nanotubes. Along with Liu, who leads the carbon nanotube densification process investigation, the research team includes associates Navdeep Bajwa, Lijie Ci, Swastik Kar and Sang Hwui Lee.

The team has filed a disclosure and is moving forward toward a patent for their densification process.

Lu’s research is supported by Defense Advanced Research Projects Agency (DARPA), the Microelectronics Advanced Research Corporation (MARCO) and the New York State Office of Science, Technology and Academic Research (NYSTAR) through the Interconnect Focus Center (IFC). The IFC is a multi-university research center focusing on the discovery and invention of new solutions that will enable the US semiconductor industry to transcend known limits on interconnects that would otherwise decelerate or halt the rate of progress toward future terascale system integration.

For more information on Lu’s research into 3-D integration, visit: www.rpi.edu/research/magazine/spring05/chips.html</br.densified>