DNA-Like Design Could Lead to Smaller, Faster Microchips

DNA may contain the building blocks of life, but IBM (NYSE: IBM) believes it can also provide the foundation for cheaper construction of the next generation of superfast, energy-efficient microprocessors.

Scientists at IBM's Almaden Research Center in San Jose, Calif., have announced they have developed a way to use synthetic DNA as a "skeleton" for the creation of microscopic circuits. The polymer-based interlocking chains of nucleic acids can actually serve as scaffolding for self-replicating nanoparticles to form miniature circuit boards.

The news has many in the microprocessor industry thinking about more economical ways of producing faster, more energy-efficient microchips, since the process of building smaller and smaller processors requires more complicated and expensive fabrication equipment. IBM and researchers at the California Institute of Technology are after a way to find loopholes in Moore's Law, which calls for a doubling of transistors that can be placed on an integrated circuit every 2 years. As the lithographic processes used to build the transitors start approaching the realm of 22 nanometers in size, researchers have to start looking at better ways of building a chip.

Blending Chemistry and Physics

One possible answer: chemistry, says Spike Narayan, manager of science and technology research at the IBM Almaden Center.

"There are no good lithography technologies beyond 22 nanometers," Narayan told TechNewsWorld. "If you want to get to 15 (nanometers), all the processes we use today simply do not extend to that. So we're on the hunt for bottom-up fabrication techniques, where we let the forces of nature assemble themselves in the manner for what we want to achieve."

IBM had already been working on technologies involving self-assembling carbon nanotubes, which can make up the switches, electrical paths and other vital components of extremely tiny processors, Narayan said. It took CalTech researcher Paul Rothemund to come up the idea of taking a DNA strand "and making origami out of it" to provide the catalyst for this latest development, Narayan added. The hope is that scientists will be able to fold the DNA into the pattern of a circuit board, and the nanoparticles will attach themselves to that pattern. The result: smaller transistors packed onto smaller, smarter, more powerful microprocessors that requires less energy to power. These will come in handy as technological advances demand more microprocessing speed and power in computers, smartphones, set-top boxes and other devices.

"The need for new materials and the need for new chemistry is starting to overtake traditional challenges in the materials and the physics space," Narayan said. "We're now entering the realm where chemistry will dominate the next-generation technologies."

The Future Is Now?

The DNA technology may be eight to 10 years away, Narayan said, and In-Stat senior analyst Tom Halfhill agrees.

"Whenever somebody says it's 10 years out, it's still basic research. The next step is production engineering, where you move everything to where it can be done in mass production," Halfhill told TechNewsWorld. "That's where some of these things can break down. You can do it in a laboratory, but then going to mass production might not work."

The possible solutions provided by the IBM and CalTech researchers intrigued Halfhill. "It's an interesting approach. DNA molecules are three-dimensional, and they're doing some folding so they can erect the circuit structure. They have to fold it down to two dimensions to make it like a regular (microprocessor) circuit."

IBM is already using some nanoscopic self-assembly techniques to create insulating layers in microprocessors, Halfhill said. His concern is that by the time the DNA technology is ready for mass production, "these DNA scaffolds might be too large," he said with a laugh. "There are already today some structures they're building into chips that are smaller than DNA -- insulation layers and transistors are averaging about four atoms thick. That's a type of silicon atom, which may be somewhat larger, but by the time they get this perfected it might be too large."

Working with carbon nanotubes might help to get around that problem, but Halfhill also says that Intel (Nasdaq: INTC) engineers believe they can break the 22-nanometer barrier using traditional lithography techniques. "From what I hear, they'll get past 22 nanometers without resorting to anything exotic," he explained.