The researchers have not built all the reading/writing circuitry needed to utilize nanoparticles as single-electron memories, but have pointed the way toward such devices by demonstrating the "scaffold" interconnection method for the test nanocomponents.

"This is the first time DNA crystals have been used to assemble nanocomponents, but it is really just the first step toward a nanoscale manufacturing scheme that can be used for electronic circuits, memories, regular periodic arrays and other things," said Richard Kiehl, a professor of electrical engineering at the University of Minnesota (Minneapolis). His research team included professor Nadrian Seeman of the chemistry department at New York University, who pioneered the use of DNA as scaffolding for nanocomponents, and Minnesota chemistry professor Karin Musier-Forsyth.

Kiehl said DNA crystals can be laid out in an array with 20-angstrom spacing to realize a memory structure with a density of 10 trillion bits/cm² — that's 100 times denser than the 64-Gbit DRAMs the electronics industry projects for 2010. However, DRAMs use random-access memory operations while Kiehl's circuits are best suited to nearest-neighbor wiring, which would require a different addressing mode.

Beyond simple memory structures, Kiehl believes the technique could serve as an interconnection breadboard for arrays of special-purpose, fine-grain processors assembled in a repeating pattern — for example, an image-processing island on an otherwise conventional silicon microprocessor chip.

"We are looking at ways to use electric fields to guide assembled DNA rafts, to mate them with electrodes that are already on the silicon chip," he said. "You can image a CMOS chip with special areas designed to contain these arrays of nanocomponents."

The group's wet-chemistry method patterns a select set of crystalline DNA molecules into tiles. The tiles have a unique sequence of chemical "hooks" along each edge and scaffolding on top to hold nanocomponents — some of which come preinstalled. The edge encoding enables the DNA tiles to self-assemble in only one way into a predefined breadboard-like scaffold, onto which nanocomponents are already anchored. The high granularity of the scaffolding method allows the nanocomponents to be placed atop a patterned silicon substrate with high precision.

"Everything is done on the silicon chip; then these islands are assembled into place," said Kiehl. "It's kind of a hybrid technology — the nanocomponents would self-assemble to gold electrodes on the surface of the CMOS chip. Since DNA is charged, we can use electric fields to guide it into place."

University of Minnnesota researchers form scaffolding for nanocomponents with DNA molecules. Lock-and-key chemistry allows self-assembly of dense arrays.

For the prototype demonstration system, nanoparticles were deposited on a 20-angstrom grid of crystalline DNA. Each gold particle could function as a single-electron storage device for a bit. In working circuits additional control circuitry would be needed to read and write bits, but Kiehl predicted that many real-world applications may want to process information locally. For instance, pixel-like cells could center on a photodetector surrounded by nanocomponents that filter, detect edges or track motion, much as the visual cortex of the human brain both senses and processes information.

"We predict that a chip made with DNA crystals and nanoparticles could perform real-time image processing for improved noise filtering, and to identify objects in images at a speed approaching that of the human eye and visual cortex," said Kiehl.

Though the team did its first demo with gold nanoparticles capable of holding an electrical charge, they pointed out that nearly any other proposed nanocomponent could be anchored to their scaffolding. Carbon nanotubes, for instance, could be used to interconnect nearby gold nanoparticles, along with any of a variety of magnetic nanocomponents that have been proposed.

"The scaffold is made by mixing together in a solution a couple of dozen different kinds of small strands of DNA that first self-assemble into tiles, and then assemble into a two-dimensional crystal," said Kiehl. "Some of the DNA strands already have nanocomponents covalently bonded to them, so once the whole thing forms, as an integral part of it you have the nanocomponents assembled into place."

Kiehl pointed out that many other researchers have experimented with the properties of gold nanoparticles, such as how to tunnel single electrons onto them so that they can be used as memory devices. But no one else, he said, has demonstrated a method of reliably laying them out in perfect arrays. By picking a set of tiles that provides different kinds of attachment points for different types of nanocomponents, Kiehl's team hopes to someday array gold nanoparticles next to carbon nanotubes and magnetic molecules that self-assemble not only memory elements, but also the processing and interconnection circuitry too.

"We are already putting nanocomponents onto chips — simple ones, but still an electronic device, because gold nanoparticles could be the basis for a single-electron tunneling device," said Kiehl.

Next up: circuit elements

For the future, Kiehl's team plans to demonstrate that its nanocomponents can function electrically as circuit elements while being mechanically bound to the DNA-crystalline scaffolding. By forming connections to and from nanocomponents held fast by the DNA crystal, the team hopes to inject signals into (and read out the results from) gold nanoparticles and similar nanocomponents.

"Next, we want to demonstrate a voltage/current measurement, showing some electrical function being performed by the nanocomponents themselves," Kiehl said. "It could be electron tunneling of an electron from nanoparticle to nanoparticle — or some other kind of electronic function performed by the nanoparticle self-assembled into the DNA crystal." Kiehl's work is sponsored by the National Science Foundation.