Improving Superconductors and Semiconductors

Dave Geohegan and Alex Puretzky use laser ablation to form carbon nanotubes for potential use in improving electronic devices. (Photo by Curtis Boles)

ORNL researchers have been developing novel materials that could lead to high-temperature superconducting wires and smaller, faster semiconductor chips that speed up data processing and flow. The materials involved include metals (ranging from nickel to silicon), ceramics, perovskite thin films, and carbon nanofibers.

COATED SUPERCONDUCTORS

High-temperature superconducting tapes made by ORNL’s rolling assisted, biaxially textured substrates (RABiTS™) technology are getting longer, stronger, and closer to commercialization.

“We have made tapes 1 centimeter wide and 50 microns thick that are as long as 2.3 meters and that carry reasonable current,” says Don Kroeger, head of the Superconducting Materials Group in ORNL’s Metals and Ceramics (M&C) Division. The RABiTS™ technology, which received an R&D 100 Award in 1999, has been licensed to the 3M Company, American Superconductor, Oxford Superconducting Technology, and MicroCoating Technologies, Inc. ORNL researchers work with these companies under cooperative research and development agreements (CRADAs).

Because RABiTS™ tapes can carry electrical current with virtually no resistance when chilled with liquid nitrogen, they will be used to make electrical devices that will take up less space, cost less to operate, and use less energy than today’s equivalent technologies. RABiTS™ wires are expected to be available commercially within the next few years for underground transmission cables, motors, transformers, and magnets.

Two recent discoveries at ORNL have improved the nickel base in RABiTS™ tapes. M&C researchers found that some alloys of nickel—with greatly reduced magnetism compared to nickel—can be textured as well as pure nickel. This is important because the magnetic character of nickel could cause significant energy losses in a conductor or device when it is operated in the presence of alternating-current magnetic fields. “At present we are getting good results with nickel with 3 atomic percent tungsten,” says Kroeger. “Its magnetism is reduced, and it is much stronger than pure nickel. At low temperature, the alloy would be nonmagnetic with 9 atomic percent tungsten.”

In ORNL’s Solid State Division (SSD), Dave Christen and Claudia Cantoni made a discovery concerning one of the buffer layers that separate the superconducting material from the nickel base. Buffer layers, such as cerium oxide (CeO₂), must be deposited on the nickel-tungsten substrate to separate it from the superconducting material, yttrium barium copper oxide (YBCO). Buffer layers prevent a destructive chemical reaction between the nickel alloy and YBCO and transfer the alloy’s texture to the YBCO, correctly aligning its crystalline units to enable it to carry high amounts of current.

Christen and Cantoni discovered that sulfur must be present to induce deposition of CeO₂ in the same crystalline orientation as the alloy’s texture so that it can be transferred accurately to the YBCO coating. “When we anneal the rolled tapes to improve their texture,” Kroeger says, “we do it in the presence of hydrogen sulfide so that sulfur is on the nickel alloy surface before the buffer layers are deposited.”

To reduce the costs of depositing YBCO on the buffer layers covering the substrate, SSD’s Ron Feenstra is experimenting with a barium fluoride technique to replace the more expensive pulsed-laser deposition process. In this newer technique, electron beams evaporate targets consisting of barium fluoride, yttrium, and copper to make a nearly amorphous coating of a chemical precursor to the superconducting phase. In a separate reaction chamber, the tape is exposed to oxygen and water vapor at a high temperature to convert this precursor to YBCO. Hydrogen fluoride is a by-product of the reaction. M&C’s Dominic Lee, Fred List, and Keith Leonard have adapted this process to produce high-performance superconducting tapes longer than one meter.

SINGLE-CRYSTAL OXIDES FOR FUTURE SEMICONDUCTORS

As transistor size is reduced to speed up data processing and flow, silicon oxide used in a transistor’s dielectric layer will be sliced so thin it will leak electrons. Shrinking transistors stop working as on-off switches or units for storing bits of information.

M&C’s Rodney McKee, Fred Walker of the University of Tennessee, and SSD’s Matt Chisholm believe they can solve this problem currently plaguing the semiconductor industry, which seeks to continue doubling annually the number of transistors packed onto each semiconductor chip. The ORNL solution is to replace the noncrystalline silicon dioxide dielectric layer with a thin film of a crystalline oxide whose superior electrical properties will allow reduction in transistor size without loss of performance.

This solution is so compelling that Motorola, Inc., is working with ORNL researchers on this new class of thin films under a CRADA. The ORNL scientists’ ability to deposit the first critical atomic layers of electrically conducting perovskite materials on semiconducting materials—strontium titanate on silicon and barium titanate on germanium—is the result of their fundamental research on the physics and chemistry of interfaces.

“We were the first to precisely control the molecular beam epitaxy process for growing thin crystalline oxide films under ultrahigh vacuum,” McKee says. “We learned how to grow a perfect film in which the film crystals are oriented in correct registry with the silicon template beneath.”

Because strontium titanate exerts a stronger influence than silicon dioxide on the transistor’s conductivity, the gate electrode in the middle that creates the electric field enabling the dielectric layer to allow or impede current flow can occupy less space. As a result, the distance the electrons must travel between the source and drain electrodes will be reduced, allowing the transistor to be made smaller and faster.

First transistor fabricated at ORNL, described in an October 2001 issue of Science magazine. The gate oxide is strontium titanate, which performs better than silicon dioxide materials, now used in metal oxide field effect transistors.

These researchers have also built a “smart transistor” consisting of a barium titanate film on a germanium substrate. This powerful transistor is smart because barium titanate’s crystal structure gives it desirable ferroelectric properties, such that in certain regions of the film, positive and negative ions separate, setting up an internal electric field that is permanent unless flipped by an external power source. As a result, the transistor “remembers” information even when the power is turned off.

A chip made of smart transistors will hold almost four times as much information as a chip today and could serve as the hard disk drive of a laptop computer, greatly extending the lifetime of laptop batteries. Crystalline oxide films on silicon or germanium substrates have the potential to revolutionize the semiconductor industry.

FINDING NEW USES FOR CARBON NANOFIBERS

ORNL researchers have grown vertically aligned carbon nanofibers in a controllable way and shown that electronic devices can be built from these nanofiber materials.

Michael Simpson, Michael Guillorn, Vladimir Merkulov, and Anatoli Meleshko, all of ORNL’s Engineering Science and Technology Division (ESTD), and SSD’s Doug Lowndes have grown patterned carpets of nanofibers on silicon substrates using plasma-enhanced chemical-vapor deposition (PECVD). They have learned to tweak the process to produce fibers having a desired position, height, diameter, shape, orientation, and even chemical composition.

This plasma-enhanced chemical vapor deposition device at ORNL is used to grow patterned carpets of carbon nanofibers on silicon substrates for possible electronic applications. (Image enhanced by Judy Neeley.)

They have also found out how to grow a needle-like carbon nanofiber that could be used as a biological probe to measure electrochemical changes in living cells or as a site-specific vehicle to deliver molecules to an individual cell. Penetrating a cell with such a “nanoneedle” would do much less damage to the cellular membrane than today’s patch clamp technique. ESTD’s Tim McKnight has demonstrated that carbon nanofibers can deliver molecules into cells without affecting cell viability. Simpson says this technology could be used for a variety of biotechnology and biomedical purposes, such as drug discovery, by using an array of nanofibers for parallel delivery of test molecules to different cells and for subsequent measurement of cell responses.

ORNL researchers have shown that carbon nanofibers can be used as field emitters of electrons that can be focused for electron beam lithography to create circuit patterns 10 to 100 nanometers across. The semiconductor industry has a goal of making chips with circuits 8 times denser and 16 times faster than the chips being produced by optical lithography today.

“We have applied a voltage using one electrode, creating an electric field that extracts negatively charged electrons from the nanofiber tip,” says ESTD’s Merkulov. “The other electrode then focuses the nanofiber’s electron beam down to a very tight spot.”

Scanning electron micrograph of a vertically aligned carbon-nanofiber-based field emission electron source with an integrated focusing electrode for applications such as electron beam lithography, which could be used to fabricate advanced semiconductor chips.

This same technology could revive vacuum tube electronics by allowing the creation of nanoscale vacuum tubes that could compete with integrated semiconductor electronics. Carbon nanofibers could also be used as vertical interconnects between transistors, especially when transistors become much smaller.

A single carbon nanofiber could be used as an atomic force microscope (AFM) probe to replace the conventional AFM cantilever. The AFM probe, which is scanned across the sample’s surface, tracks the surface morphology in response to atomic forces between the cantilever tip and the surface. The conventional pyramid-shaped tip is too big to fit inside surface “trenches,” but a long carbon nanofiber “needle” could easily drop into a trench and scan its interior.

Carbon nanofibers resemble stacks of funnels or cones and have a crystalline structure different from that of carbon nanotubes, which are hollow cylinders made of six-member carbon rings. ORNL researchers are studying the fibers’ chemical, electrical, and mechanical properties, which are expected to be different from those of carbon nanotubes.

The process used at ORNL to grow nanofibers is to deposit catalytic seeds—dots of nickel, cobalt, or iron—in a desired pattern on a silicon substrate. The substrate with an array of metal dots is then placed in the PECVD furnace. Acetylene (C₂H₂) and ammonia (NH₃) are introduced into the furnace, which is heated to 700°C. When power is sent to two electrodes in the furnace, a plasma is formed between the electrodes, setting up an electric field.

“The electric field from the plasma is needed to make the nanofibers grow aligned, pointing in the same direction,” Merkulov says. “The acetylene is needed to supply carbon for growing nanofibers on the metal dots. We found the ammonia is needed as an etching gas to keep the acetylene plasma from covering the cathode with carbon, which would kill the process. The ammonia also supplies nitrogen that can get into the carbon nanofibers. We are studying how nitrogen affects nanofiber properties.”

The ORNL group was the first to develop a good understanding of the PECVD process (invented in 1998 at the State University of New York at Buffalo) and the first to apply this knowledge to the fabrication of field-emission devices using carbon nanofibers.

ORNL researchers have also shown that the carbon nanofiber production process using PECVD and standard microfabrication steps taken at the Cornell Nanofabrication Facility can be used to build identical electronic devices on a large scale, with all parts of the devices arranged in a desired geometry and with input and output capabilities.

Although nanofibers are very thin and short, their effect on future biological and electronic technologies could be far reaching.

Beginning of Article

Related Web sites

• ORNL's Metals and Ceramics Division
• ORNL's Solid State Division
• RABiTS™

Search Magazine