UCLA Chemists, Hewlett-Packard Labs Colleagues Report Significant Advances Toward Chemical Computers

In the July 16 issue of the journal Science, Heath and his colleagues demonstrate molecular-based logic gates for the first time and show that, for certain tasks, molecules can effectively achieve the same or better results than silicon. They also show that molecular circuitry can be defect-tolerant.

"What we have here for the first time is a molecular device that is a real technology -- not just an isolated device," Heath said. "This is a real step toward making a molecular computer."

Molecular computers hold the promise of being far less expensive and much smaller and faster than today's silicon-based computers, and to learn and improve the more they are used, Heath said. Molecular-electronic-based computers would also have vastly reduced power consumption. In addition, such computers would contain vast amounts of resources; this implies that all data could be securely hardwired into the machine, and that nothing would ever need to be erased -- making such machines immune to disruptions such as those caused by computer viruses.

"In order to make a computer, all you really need are wires and switches," Heath said.

The research team also includes J. Fraser Stoddart, UCLA's Saul Winstein Professor of Organic Chemistry; Francisco Raymo, a postdoctoral scholar in Stoddart's laboratory; Martin Belohradsky, a former postdoctoral scholar in Stoddart's laboratory; Philip Kuekes, a researcher in the Computer Systems Lab at HP Labs; and Stan Williams, director of the Quantum Structures Research Initiative at HP Labs.

How much improvement can be made over today's computers that run on silicon semiconductors?

"You can potentially do approximately 100 billion times better than a current Pentium in terms of the energy required to do a calculation," Heath said. "When you look at a technology that can be improved to such an enormous extent, you know someone is going to do it. Silicon's energy efficiency is not going to improve by much more than a factor of 10. I believe we can improve energy efficiency by at least six or seven orders of magnitude -- not just make silicon a little bit better, but move into a realm that silicon could never achieve."

From science fiction to actual science

Heath believes there could be prototypes of molecular-electronic-based computers in just a few years, and a hybrid computer with substantial molecular electronics and a small amount of silicon-based technology in perhaps a decade.

"What once seemed like science fiction is now looking more and more like actual science," Heath said. "We can potentially get the computational power of 100 workstations on the size of a grain of sand. We'll do it in steps; I'm hopeful that we can do it in about a decade. Years ago, when I first told people I was trying to make a computer chemically, I wasn't taken very seriously. Although the idea is getting a little more respect now, we still have a long way to go."

The class of molecules the chemists are using are called rotaxanes -- synthetic, dumbbell-shaped compounds that were first invented in Stoddart's lab.

"We use rotaxanes as molecular switches, but the key is how they are used in an architecture that is structurally simple, but logically complex," Heath said. "The molecular switches by themselves are useless, the wires by themselves are useless, but the architecture is important. We should be able to configure these wires and switches to do what a very complicated silicon-based circuit does -- including performing logic operations, providing memory, and routing signals through the machine and to the outside world."

To make a molecular computer, the first critical step is to take a set of wires arranged in one direction, a layer of molecular switches, and a second set of wires aligned in the opposite direction. At the junction of the wires is a single layer of molecules -- the rotaxanes. The chemists electronically configured these wires and switches to fabricate logic gates. They showed they could link molecular switches and wires together and reconfigure the logic circuit as needed.

The rotaxanes, which Stoddart provided, worked better as switches than Heath had expected, and allowed Heath's team to configure wire-switch networks into highly effective logic gates.

Many severe challenges remain, but Heath is guardedly optimistic.

"Chemistry and silicon-processing technology are almost incompatible with each other," he noted. "When you make a computer using silicon-based fabrication methods, you make something that looks complicated and is perfect. Anything you make chemically is not complicated and inevitably has defects. However, we can take a set of wires and switches with defective components, find the defects and route around them. In the lab, we can make pretty good components already. We'll get a lot better. We've just begun."

The research was funded by the Defense Advanced Research Projects Agency, the National Science Foundation and the Office of Naval Research.

Heath, who works in nanotechnology, said the ultimate challenge of his field is to develop a true, three-dimensional, manufacturing technology, which currently does not exist.

"Anything that is manufactured -- high-tech or low-tech -- is either one-dimensional or two-dimensional," he said. "Biological systems are three-dimensional, but all our manufacturing is not. If we can learn how to make this molecular-electronic computer, we will take a long step toward three-dimensional manufacturing. Then you can imagine revolutionizing many things, from industry to medicine."