Technology News
06/01/2004
Transistor nodes along carbon nanotube interconnect pipes
Fujitsu researchers are working on making simple transistor nodes along networks of carbon nanotube interconnect pipes. This opens the possibility of "fundamentally changing the nature of the IC," says Naoki Yokoyama, Fujitsu Research Fellow and general manager of the company's Nanotechnology Research Center. Instead of making transistors on the wafer and then joining them with interconnects, chipmakers would make transistors directly along the interconnects.
Figure 1. Fujitsu researchers are working on an IC using carbon nanotubes that can act as either metals or semiconductors. (Source: Fujitsu, Nikkei Microdevices) |
One form of carbon nanotube acts like a semiconductor, with electron mobility 10× that of silicon. But with a different atomic structure, a carbon nanotube can act like a metal, with current density 1000× that of copper. Researchers make a metal-type nanotube around the outside of a semiconductor-type, then form source, gate, and drain electrode rings around the double tube in places. Running current between units then burns out the outer metal layer between those electrodes, exposing the inner semiconductor layer to make the transistor, creating a simple FET right along the interconnect (Figs. 1 and 2). Such devices are still just concepts, though Fujitsu says it has developed technology to selectively expose the semiconductor layer.
Figure 2. Schematic of an FET made from carbon nanotubes. (Source: Fujitsu, Nikkei Microdevices) |
Work is also progressing on using nanotubes as more standard IC interconnects, for tiny, high current-density vias. Fujitsu researchers deposit Ni at the bottom of the via holes as a catalyst, then use a CVD process to fill the holes with a mass of vertical carbon nanotubes. The protruding ends of the nanotubes are smoothed off by CMP, conveniently removing the high-resistance end part of the nanotube (Fig. 3). Using very fine particles in the Ni catalyst increases the density of the nanotube growth and further lowers resistance, though it still needs to be brought down even more to be practical. The company is aiming for applications by 2010.
Figure 3. Vias made with carbon nanotubes. The ends of the tubes are removed by CMP to decrease resistance. (Source: Fujitsu, Nikkei Microdevices) |
Masahide Kimura
Nikkei Microdevices, SST partner
Light and sound from nanocrystalline silicon
Professor Nobuyoshi Koshida's group at Tokyo U. of Agriculture and Technology sees wide potential for displays and actuators from its process of getting planar light and ultrasonic emissions out of nanocrystalline silicon.
Treating polysilicon with anodic oxidation to make crystals 2–3nm dia. creates a material that can be made to emit ballistic electrons at atmospheric pressure, and ultrasound.
The anodization process leaves each nanocrystal coated with SiO2 film. When voltage is applied, a strong electrical field is created at the crystal boundaries and some accelerated electrons are emitted. Since the crystals are so tiny, the electrons pass through the solid with very little scattering, getting accelerated at each crystal boundary in a chain reaction until they are emitted from the surface, creating fluorescent light. This ballistic emission, moreover, is from the entire planar surface, not just from points at the ends of needle-like elements as with other light emission technologies, so it could create a much more uniform display. Also, the electrons are accelerated within the solid, so the process works at atmospheric pressure, with no vacuum required.
Researchers are working on developing a display using the technology. So far the emitted light is still dim, but they figure that making the structure of the nanocrystalline silicon more regular, and reducing the defects at the crystal boundaries, will improve acceleration of the electrons to make brighter light.
The nanocrystalline silicon can also be made to emit ultrasound. Researchers grow a 10–50µm layer of the nanocrystalline silicon on a conventional single-crystal silicon substrate, then put metal circuits on top. Because the nanocrystalline silicon is highly insulating, heat applied to its surface dissipates into the air above instead of the substrate below.
When alternating current is sent through the circuits, Joule heating of the surface of the nanocrystalline layer comes from alternating pulses. These quick changes in surface temperature expand the air and change the pressure above, generating waves of ultrasound. The technique can theoretically generate ultrasonic waves of up to 1GHz, without physical motion. Researchers figure the impulse waves can be used in a wide range of noncontact actuators (Figs. 1 and 2).
Masahide Kimura
Nikkei Microdevices, SST partner
Sensors, algorithms speed EFEM robot setup time
Sometimes it takes more than a day to set up and "train" the precise movement of wafer-handling robots in semiconductor fabrication equipment frontend modules (EFEMs), which are bolted on fab tools to load and unload wafer carriers in protective atmospheric mini-environments.
Various levels of software and automatic controls have been introduced for calibrating EFEM robots, but now Newport Corp. says it is taking "self-teaching" further by combining patent-pending learning algorithms, transmissive optical sensors, and robotic hardware in its first EFEM, called Performix.
In the "Automated Self-Teach" setup mode, Performix robots use LED-based optical sensors on end effectors to "scan up and down and left and right to determine location, and then it locks in that information," explains Ed Dante, senior product manager for automation at Newport's Robotic Automation Division. "It teaches itself in tens of minutes as opposed to several hours or a couple days, in some cases, with traditional approaches."
Competing systems are also aiming to ease the job of calibrating and teaching robots where to move — and not. For example, Berkeley Process Control Inc.'s Autocalibration technology uses software to sense changes in torque and velocity of robotic arms for real-time control and setup. New capabilities, such as in situ diagnostics, are being added to Berkeley's software and BX-series of motion-and-machine controllers aimed a range of robotic hardware in fab equipment and EFEMs.
But Newport is branching out into the EFEM arena, offering complete self-teaching subsystems to compete with the likes of Brooks Automation Inc. and Asyst Technologies Inc. The Performix can achieve throughputs of 200 wafers/hr, and versions are aimed at 200mm SMIF as well as 300mm FOUP carriers. The system will debut at Semicon West, and Newport expects its first EFEMs to be deployed in a fab during the third quarter, says Dante.
Coming up next on Newport's roadmap are autoteaching wafer stages for metrology equipment. Look for those subsystems to appear by 4Q04, if not sooner. — J.R.L
Researchers boost the efficiency of polymer OLEDs
A presentation at the American Chemical Society's recent national meeting suggests that the designers of flexible displays made from polymer organic light-emitting diodes (OLEDs) should be able to boost the efficiency of the systems beyond what most researchers had expected.
Simple statistics based on electronic spin theory suggest that only 25% of the energy put into the displays will produce light, but careful manipulation of the polymer chains could increase that efficiency to 50%, said Jean-Luc Brédas, a professor in the School of Chemistry and Biochemistry at Georgia Institute of Technology. Polymer LEDs consist of a thin film (0.1µm) of a polymer such as polyparaphenylene vinylene sandwiched between two electrodes. They are usually built on a transparent substrate that can be glass or flexible plastic.
When voltage is applied to the electrodes, the cathode injects electrons into the polymer film, while the anode injects positive charges, also known as holes. Those charges migrate along the polymer chains until they meet. When the charges meet, a two-step charge-recombination process takes place in which the opposite charges neutralize one another, producing an excited state (an exciton) in the polymer. The decay of that excited state is what can produce light.
By taking advantage of complex restrictions on the amount of energy that can be released by the materials during the recombination process, Brédas and his colleagues show theoretically that systems built from long polymer chains should be able to boost the percentage of light-emitting singlets to as high as 50%.
The Georgia Tech research has been supported by the US National Science Foundation, the Office of Naval Research, and the IBM Shared University Research Program. The other universities involved are the U. of Mons-Hainaut in Belgium and the Chinese Academy of Sciences.