The "circuit" that the researchers have designed is a complex molecule based on the polymer polyphenylene. The molecule uses molecular wires and molecular diodes that have been demonstrated in the lab. "All of these detailed logic designs include only experimentally demonstrated molecular electronic devices as their components," the researchers said in a report. The electronic half adder accepts one-bit binary inputs and has one-bit outputs. The molecule computes the sum and the carry bits. "The corresponding conductive molecules, if realized, would use much less power and also would be as much as one million times smaller in area than the comparable circuits on a state-of-the-art commercial microcomputer chip," according to the report.

The field began with the realization that long chain molecules could serve as conductors. In a conventional wire, electrons drift through the material in bulk, creating a current. In a molecular wire, the electrons hop from one molecule to the next in the chain. Once conduction through such a wire has been demonstrated, the next step is the creation of gating devices. The Mitre researchers have been able to use a promising technique for the field to create rectifying diodes.

The key technique is the ability to introduce "dopants" — similar to the p- and n-type dopants used to build silicon devices — into the molecular equation. That was accomplished by attaching side chains to a molecular wire. By first attaching a molecule that "donates" electrons and then one that "accepts" or removes electrons, it is possible to build molecular-scale rectifying junctions. This is only the most primitive kind of electronic component, but it is where the semiconductor business started. The first semiconducting device built at Bell Labs used germanium as the medium, and dopants were introduced to play a similar role of either adding or subtracting electrons.

The next goal of the research at Mitre is to find a chemical recipe for a three-terminal transistor, opening the door to truly complex logic devices. In addition to small size, such components could benefit from the automatic assembly processes of chemistry. Unlike attempts to build electronic devices using nanoscale manipulators like atomic force microscopes, a chemical approach to circuit synthesis could lead to the building of immensely complex devices using chemical reactions as the means of fabrication.

Defects downside

The one drawback to that scenario is the problem of random defects. Even as conventional semiconductor processes reach down into smaller scales, the problem of physical defects poses formidable barriers. While molecular structures can self-assemble with a high degree of accuracy compared with human-directed processes, the sheer number of units means that even a small probability of error will result in many defects. And, at the molecular scale, a defect can break a connection or render a junction inoperable. With state-of-the-art 0.18-micron processes, the defect problem is troublesome, but it can be contained.

For molecular electronics, the question of how defects will be handled is an important design constraint. Some groups are working with computer models of fault-tolerant circuit designs. The idea is to find circuit architectures that will be able to function accurately even with a given level of random defects. Biological systems have been able to solve that problem using essentially parallel redundant networks.

The neurons in the brain, another spontaneously assembled circuit, are able to function well despite highly unreliable components. Neurons frequently die or operate abnormally, and yet the brain is able to continue as though nothing were wrong. Similarly, researchers using the molecular circuit design approach will be able to take advantage of a large number of circuit components and interconnects, since the circuits will spontaneously assemble out of a staggering number of molecular components.

The challenge for chemical researchers is to find specific reactions that can be guided to create the logic operations needed for computers that will operate even with a high level of defects.