What exactly is a molecular switch?

It’s a molecule that shows bistability. In other words, if you put a molecule between two wires and try to run a current through it, the only way current gets through is by tunneling. And because tunneling is a quantum-mechanical process, anything you do to this molecule exponentially affects tunneling. If you take this molecule and, for instance, it’s in one structural isomer and somehow you get it to move to a different isomer, then you will change the current through this device, and that’s the basis of a switch. The configurable bits used in Teramac have a memory bit that sets to zero or one, which in turn opens or closes a switch, and the whole thing has six transistors and four wires. A huge amount of overhead. That’s why these configurable bits aren’t used in all computers. So if you could get a good molecular switch, you could replace all this with two wires and a molecule. It's a big deal. The one thing you don’t get is gain, but it turns out that that’s not so important, because you can probably feed that in through conventional silicon transfers.

So we started working very hard on taking small lattices of these components, which have wires and molecules in between them, developing a legitimate molecular switch, and then electronically turning this into something that will do a logic function. Last summer we published an example of doing this, which was just an architecture demonstration, not a technology demonstration. We had molecular switches but we could only switch them once. But that’s how we ended up on the front page of the New York Times.

High-Impact Papers by James R. Heath, Published since 1990 (Ranked by average citations per year) Rank Paper Total Citations Average cites per year 1 C.P. Collier, et al., Reversible tuning of silver quantum-dot monolayers through the metal insulator transition, Science, 277(5534):1978-81, 1997. 80 23 2 D.V. Leff, et al., Thermodynamic control of gold nanocrystal size: Experiment and theory, J. Phys. Chem., 99(18):7036-41 , 1995. 105 23 3 J.R. Heath, C.M. Knobler, D.V. Leff, Pressure/temperature phase diagrams and superlattices of organically functionalized metal nanocrystal monolayers: The influence of particle size, size distribution, and surface passivant, J. Phys. Chem. B, 101(2):189-97 1997. 76 22 4 J.M. Hawkins, et al., Organic chemistry of C60 (buckminsterfullerene): Chromotography and osmylation, J. Org. Chem., 55(26):6250-2, 1990. 136 14 5 D.V. Leff, L. Brandt, J.R. Heath, Synthesis and characterization of hydrophobic, organically-soluble gold nanocrystals functionalized with primary amines, Langmuir, 12(20):4723-30 1996. 47 13 6 P.C. O'Hara, et al., Crystallization of opals from polydisperse nanoparticles, Phys Rev. Lett., 75(19):3466-9, 1995. 68 12 7 A.A. Guzelian, et al., "Synthesis of size-selected, surfacepassivated INP nanocrystals," J. Phys. Chem. B, 100(17):7212-9, 1995 53 12 SOURCE: ISI's Personal Citation Report, 1981 - June 2000

Meaning you had only one chance to take the system and turn it into a logic circuit? But isn’t that how silicon works?

Yes. But what we would like is a system in which we can just download whatever structure we want onto it. So we’ve spent a lot of time working on reconfigurable switches, and now we have four. And they’re all very different. Two of the switches can be switched electronically, another one chemically, and another optically. But they’re all based on the same exact architecture. Molecules just do things like that. This was work done primarily with Fraser Stoddart, but also with Fred Wudl--both UCLA chemists.

So how do you turn those into a working computer?

What we’re going to make are two-dimensional crystals with nanotubes or silicon nanowires. You put down a two-dimensional crystal of wires–all pointing in one direction. Put molecules down everywhere–just one monolayer of molecules. Very carefully making sure it’s just one monolayer. And then you put down the next set of wires, in the other direction, and now you have to interface all these wires. That’s the entire manufacturing process for this computer, except that you then have to electronically download the logic structure on these resources. But that’s everything in the computer.

What are the long-term prospects for this technology?

Well, basically if you look at silicon-based manufacturing, it's becoming a very mature technology. For lots of reasons, one of which is that it’s based on a solid-state switch, a transistor, and for fundamental physical reasons, it's getting very expensive to make improvements. If you look at the major players in the chip business, they’ve been dropping like flies in the past several years. It’s really now just IBM, Intel and Motorola in the U.S.

But more than that, Richard Feynman and IBM’s Rolf Landauer calculated the ultimate thermodynamic limit for the efficiency of a computing device and it’s about 10¹⁸ operations per second at one one watt of power consumption. That is a factor of a billion better than any machine silicon is ever going to produce. This says that even though silicon may have reached its maturity, computation is just beginning. You can do a lot better. But to get down to this thermodynamic limit, you have to have a quantum state switch, which means just one electron determines what’s zero or one. And it has to work at room temperature. It can’t be some exotic liquid-helium-cooled thing. This implies that it can be no bigger than 100 angstroms. So you already have a length-scale that’s molecular. And molecules do this as well. Having a bottom-up-assembly, quantum-state-switch-based machine is probably the only way you’re going to get there. And this 10¹⁸ issue means there is a very, very strong motivation to do it.

What about competition? There must be a lot of research labs working on this.

A number of corporate labs have started molecular-electronics programs just within the past year or so. More generally, however, there is this big field of nano, and while people out there are making significant headway in fabricating very small devices and understanding their properties, only a few people are trying to think about building an entire machine. However, some folks are beginning to appreciate that if you put all of this stuff together, then you might make a machine that really does something unique and great. It's very rare in any physical science that you see a figure-of-merit improvement of a billion that's just waiting to be tapped.