Vivek Subramanian, EECS Dept. UC Berkeley

Alex introduced his work on carbon and boron nitride nanotubes. Since nanotubes are formed via sp2 bonding (which is the strongest bond known), they have excellent mechanical properties. Alex talked about the various types of carbon nanotubes available — metallic, small bandgap semiconductor, and large bandgap semiconductor. In general, the semiconducting nanotubes are p-type, though they can be made n-type by doping with potassium.

Alex’s group has demonstrated the formation of schottky barriers between metallic and carbon tubes. In fact, he has demonstrated that they can be formed within a single tube containing a defect that alters the chirality along the tube (i.e., one side metallic, the other side semiconductor).

Carbon nanotubes can also be used as gas sensors. For example, O₂ adsorbates have been shown to dramatically alter the electrical characteristics of the tubes. They may potentially be used to form very sensitive gas detectors.

In addition to working with single-wall nanotubes, Alex’s group has used multiwall nanotubes to demonstrate numerous interesting machines, including a variety of spring structures, etc.

At the end of his talk, Alex talked about his more recent work on Boron nitride nanotubes. These may be interesting, since they do not show a change in electrical properties with chirality.

Jeff discussed his work on integrating carbon nanotubes with CMOS. His group is currently working on prototyping a decoder to enable rapid characterization of hundreds or thousands of tubes deposited by CVD on a CMOS substrate. They hope to use this as a tool to enable statistical studies of tube characteristics, leading to guidelines for optimization of chirality, etc.

Philip discussed IBM’s work on carbon nanotube FETs. He showed the progression in results that has occurred at IBM, starting with the initial FETs that had very high contact resistance. Titanium carbide was used to reduce the contact resistance, and a top gate was used to improve the electrostatics of the devices. The result was a FET with S <130mV decade, and a transconductance >1000uS/um, which is better than that achieved using a silicon FET of the same geometry.

Work still remains to be done, since even the current FETs are contact limited. Besides the challenges at the individual device level, an integration technology remains to be developed. Philip likes the chances for CNT-devices, since they "look and feel" a lot like silicon FETs.

Several interesting comments resulted from discussions during the various breaks. All speakers felt that the variability in nanotubes was the single biggest roadblock to their use in large-scale circuits. For example, separating metallic and semiconductor nanotubes will be necessary before they are used in large circuits (perhaps with semiconductor CNT FETs and metallic CNT wires). Alex was of the opinion that liquid-phase self-assembly was the likely method of achieving this (i.e., the various grown tubes could be presorted and then deposited in the appropriate pattern onto a wafer using some sort of self-assembly process). Alex also felt that BN was interesting in this regard, since it has a chirality independent bandgap. Jeff pointed out that some of the first "usable" applications of CNTs may be in mechanical structures, rather than electrical ones, and people need to spend more time looking at mechanical device integration issues.

Kris talked about his group’s work on Smart Dust. He highlighted the progress on the development of TinyOS, an operating system <200 bytes in size, already being looked at for applications beyond Smart Dust. Kris showed how Smart Dust motes had been used for seismic monitoring and for power monitoring during the power crisis last year. He then went on to discuss the scalability of the motes, specifically in terms of energy. For energy reasons, communication between motes currently occurs using modulated optical reflection of an incident laser beam, since this enables communication at an energy cost of ~2pJ/bit, vs., ~1nJ/bit required for RF communication. However, Kris discussed his plan to introduce RF-based motes in the near future.

Hans discussed technologies that could change the "Moore’s law" dynamics. He introduced the work at IBM on quantum computing, and illustrated that recently, NMR was used to perform a simple factoring calculation. He then went on to discuss the future of storage, and talking about the scaling limits of hard drive technology, analogous to the more familiar silicon scaling limits. For example, the flying height of the drive head over the platter is near a limit, analogous to the gate oxide scaling limit in Si. Has then discussed the potential use of arrays AFM tips to perform parallel reads. He generally felt that self-assembly phenomena were the key to many of these "future" technologies.

Simone discussed her work on nanostructured materials for magnetic recording applications. She discussed her work on the self-assembly of cobalt nanoparticles in a polymer matrix. Cobalt nanoparticles were synthesized in solution and then assembled in the polymer. Using this scheme, a sigma of 0.05 was achieved, vs. a sigma of 0.6 for conventional "grains" in blanket deposited films. Assuming one bit could be stored per particle, this could potential result in a storage density of 10Tbit/cm2.

She also briefly discussed other nanoscale patterning technologies being studied at IBM, including the use of diblock copolymers, etc.