Portland, Ore. Just as a poem is composed from a toolbox of 26 characters or a fugue from a 12-tone scale, molecular-scale chips will one day be built from the four-letter alphabet of DNA, a Duke University EE professor believes.
To prove the point, Chris Dwyer and his collaborator, Duke professor Thom LaBean, recently demonstrated how to pattern 100 trillion 16-cell (4 x 4) programmable arrays with 2- to 10-nanometer features compared with 65-nm features today using just $40 worth of commercially available dioxyribonucleic acid.
"We are demonstrating the first steps toward using DNA self- assembly to organize circuitry at the scale that chips will be using in 10 to 20 years," said Dwyer. "We are creating, with proteins, aperiodic patterns that will be useful for circuitry. Ultimately they will be populated with electrically active nanoparticles and optically active molecules. We have plans for both."
The 4 x 4 grids in the demonstration were programmed to spell out "DNA," but the researchers claim that eventually an entire memory bit cell could be programmed into larger versions of the grids. Once the grids are populated with the necessary components to form programmable bit cells, they could be deposited on silicon substrates as thin films, or they could potentially be left in the solution in which they were programmed and then addressed with a scanning laser.
"Leaving the components in solution would introduce a new level of randomness into random access, because you wouldn't know just where to scan to find any particular memory cell," said Dwyer.
The circuitry realized on the grids could have optical qualities too. For instance, the circuits might yield liquid-crystal displays with floating pixels that could be toggled to the correct color, depending on where the pixels happened to be floating at any particular moment.
"What we have created is a breadboard, so to speak, where we can do selective chemistry independently in each of the 16 intersections of a 4 x 4 grid. We can deposit different nanoparticles and do all sorts of chemistry at each of those locations, and in that way control the pattern of the circuitry," said Dwyer.
Each of the 16 tiles comprised nine coded strands of DNA, obtained from a DNA sample that the researchers had purchased commercially. By performing various wet-chemistry steps in the lab because DNA can be damaged by direct exposure to air the researchers were able to coax the floating DNA strands to self-assemble into the grids of 16 tiles, each coded for attaching an electronic nanoparticle, an optical molecule or a nanowire connection.
"We mix them together, and they self-assemble into 1,014 of the little 4 x 4 grids, which you can think of as templates," Dwyer said. "They essentially form a sacrificial scaffold for the templated materials."
The upside of using the simple, four-letter DNA "alphabet" of proteins for creating templates is the ease with which almost any combination of self-assembling components can be made. Here, the researchers began with strands of DNA coded in their center to attach to like-coded strands to form crosses, which in turn were coded with sticky ends that enabled self-assembly into the grids. The middle of each cross was coded to accept a different electrical or optical component as appropriate to the circuit.
The downside of so many combinations was that an enormous amount of computer time was required just to code the strands to spell out DNA. "We found out there are a lot more combinations than you would think," said Dwyer. "We had to run searches on 300 computers for over two weeks."
Duke's Office of Licensing and Ventures is considering applying for a patent on the technique, even before the researchers have produced a single circuit which is their next goal.
"Next we want to design even larger grids ones big enough that we can begin to self-assemble working circuits," Dwyer said. The researchers have not yet decided whether to aim for a simple logic circuit, an analog sensor or an optical function for their first working circuit.
Working on the project with Dwyer and LaBean were Duke professors John Reif and Alvin Lebeck; researcher Sang Jung Ahn at the Korea Research Institute of Standards and Science; and Constantin Pistol and Sung Ha Park, who at the time were both graduate students at Duke (Park is now a postdoctoral fellow at the California Institute of Technology).