An "ultra-low consumption" molecular switch

Paris, February 25, 2003

The energy necessary to operate an ON-OFF molecular switch using a single molecule has just been determined by means of an atomic force microscope. To ensure switching, researchers limited themselves to the conformation changes of a chemical bond controlled by the tip of the microscope. The energy corresponding to the operation of the molecular switch is therefore 10,000 times lower than that of a microelectronic transistor. This research was carried out by the nanoscience group of the Materials Manufacturing and Structural Studies Center (CEMES-CNRS, Toulouse, France), in partnership with the Physics Department at the University of Basel in Switzerland and IBM Zurich. Its aim is to replace individual transistors with a single molecule within a molecular electronics framework.

Two years ago, the Physics Department of the Free University of Berlin in Germany, in partnership with the nanoscience group of the Materials Manufacturing and Structural Studies Center (CEMES-CNRS, Toulouse), developed a molecular switch on a cyclic molecule of porphyrin substituted with four di-butyl-phenyl groups. Of these four molecular legs, three were used to form a stable tripod for the porphyrin. The researchers then placed the molecule on a copper surface acting as a contact electrode. In order to obtain a molecular switch, they only had to create an intramolecular switching effect by applying the tip of a scanning-tunneling microscope to the remaining group. The rotation of the phenyl around its junction with the porphyrin results in only two stable positions: one that is perpendicular to the copper surface and the other that is parallel. Because one of these allows for high conductance at the metal-molecule-metal junction and the other does not, the two molecular conformations formed the "on" and "off" positions of the switch.

This new research, presented in the February 14, 2003 issue of Physical Review Letters, made it possible to determine the switching energy consumed by this new electronic component. This measurement has never been determined and could not have been made without using a new version of the atomic force microscope ultrahigh vacuum, in partnership with IBM Zurich. Researchers worked by applying constant oscillation to the cantilever above the phenyl and by recording frequency variations of the cantilever according to the vertical position of the tip. The junction between the tip of the microscope and the surface of the molecule thus resulted in the switching of the phenyl and made it possible to deduce the energy required for the switching. The CNRS team in Toulouse used these results as a basis for modeling the experiment on the computer and calculating the switching energy corresponding to the measured force. This value was calculated at 47 zeptoJoules (47 x 10^-21 Joules), a lower value than that of the theoretical rotation energy of a carbon-carbon bond. Molecular modeling solved this problem by showing that the switching of phenyl is preceded by a distortion of the overall molecule due to the tip that limits the quantity of energy required for this switching.

Even if this research represents a true technological feat and takes scientists to the limits of thermodynamics, the future of these new molecular components is still not a certainty. With consumption of the order of 50zJ, or 10,000 times less than a microelectronic transistor, this so-called hybrid approach would have seemed to be a sure thing, interconnecting molecular components and metal wires. But as Christian Joachim has pointed out, the miniaturization of microelectronics has already come up against a certain number of obstacles at this time. "If we combine the forecasts related to Moore's law (more and more components per chip) and the evolution of the consumption of components that cannot break free of thermodynamic laws, we will have chips using almost a megawatt of power each in ten years from now." The synthesis of a single molecule encompassing the entire range of electronic functions would then represent a viable alternative.

References: