John Timmer at Ars has a wonderfully detailed write up of a bit of research just published that confirms the possibility of shrinking the fundamental element of modern computing and electronics down to the scale of a single molecule.
The tour de force in the work, however, was the fact that the authors imaged the molecule sitting within the gate, and tracked the effect of the changes in gate voltage. A technique called inelastic electron tunneling can detect the vibrational modes available to the atoms in the molecule; the researchers used it to demonstrate that applying a voltage to the gate changes the energy of the orbitals, with a corresponding impact on the vibrational energy.
As an accompanying perspective points out, this is far from a trivial confirmation of the expected. Because of the size of the objects involved in producing molecular-scale gates and the high failure rate of their production, there’s a very real risk of researchers being led astray by a contaminant or malformed structure of some sort that ends up acting like a gate. By demonstrating they could detect the actual molecule in question responding to the gate voltage, the new results leave little room for doubt.
The remaining challenges to solve related to produce the transistors with a high degree of reliability. If it takes a thousand attempts to produce a single such element, then it just isn’t going to scale very effectively to the point where we can even contemplate commercial applications. As Timmer points out, the other challenge is that the transistor is only one part of a computer chip or electronic circuit. Improvements still need to be made around the transistor to reap the full benefit of being able to push circuit density completely down to the single molecule scale.
I also have to wonder what problems of scale will start to arise as this research progresses. At measurements of tens of nanometers, our gadgets already start to suffer from strange effects and interference from the realm of quantum mechanics. Does the very nature of the benzene molecule used here avoid those same problems or will molecular transistors have to be bogged down with even more error correction and compensation apparatus?