The Behavior of Molecules as Electronic Devices

he power of computer processors has grown by leaps and bounds as scientists and engineers have found new ways to fit more and more microscopic electronic devices, or transistors, onto silicon chips. Today's chips can hold tens of millions of these transistors. However, the density of silicon-based transistors on a chip is limited by the laws of physics, and engineers are fast approaching these limits. Sokrates Pantelides and colleagues at Vanderbilt University are using NPACI's high-end computing resources to study how individual molecules might be used to replace silicon-based transistors. No one has yet created a molecule-based transistor, but Pantelides' models are revealing the molecular behavior that engineers will one day use in building such nanodevices. The result may be computer chips with 10,000 times more transistors in the same amount of space.

BUILDING TRANSISTORS FROM MOLECULES

APPLYING FULL QUANTUM MECHANICS

HANDLES FOR ENGINEERS

The computational work of Sokrates Pantelides and Massimiliano Di Ventra of Vanderbilt University is solving quantum physics equations to examine a form of benzene as a two-terminal nanodevice. This image, provided by Mark Reed of Yale University, shows benzene's six carbon atoms (colored spheres), sulfur atoms (gold spheres), and electrical contacts (rough gold surfaces).

BUILDING TRANSISTORS FROM MOLECULES

The smallest electronic building blocks can be described generally as two-terminal or three-terminal devices. A light bulb is an example of a two-terminal device. With each terminal on the light bulb connected to opposite ends of a battery, current flows through the bulb. A three-terminal device has two terminals for creating an electronic circuit, and a third terminal that acts as a switch. Transistors in a computer chip are three-terminal devices that are used either as on-off switches or as amplifiers.

Building a molecular transistor first requires understanding how a molecule behaves as a two-terminal device--in other words, as a "wire" in an electronic circuit. "You have to know how current passes through the molecule," Pantelides said. "How it behaves as you increase the voltage is key to using the molecule in a device." This property, the current-voltage characteristic, measures how the current through a molecule changes as the voltage increases.

Funding from DARPA and the Office of Naval Research enabled Pantelides to hire postdoctoral researcher Massimiliano Di Ventra. This funding also leveraged an IBM grant with which Pantelides acquired a 32-processor IBM RS/6000 SP. Di Ventra has led the group's computational work to solve the relevant equations of quantum physics and examine the current-voltage characteristics of small molecules, such as a benzene ring, as two-terminal systems. Such current-voltage characteristics were measured recently by James Tour at Rice University and Mark Reed at Yale University (Figure 1).

Pantelides' group then added a third terminal to the benzene molecule to analyze how the third terminal affects the current-voltage characteristic. The third terminal is analogous to a "pinch" in a length of garden hose through which water is flowing. No pinch lets the current flow unhindered--the on or '1' state in the transistor--while a hard pinch blocks the current--the off or '0' state. And just as partially blocking the flow of water through a hose can increase the force of the flow, a partial pinch on the molecule can amplify the current.

"No one has fabricated a three-terminal molecular device in the real world, but you can model it," Pantelides said. "Our simulations allow us to ask what governs the ups and downs of current as voltage increases. This property is well known for silicon devices, but not for molecules." Simulating a three-terminal molecular device will guide researchers toward successfully building these devices.

Experimental current-voltage characteristic of a benzene-1,4-dithiolate molecule as measured by Reed, et al. (top). Two theoretical curves corresponding to the benzene molecule attached directly to flat electrodes (middle) and to single gold atoms protruding from a surface (bottom). The extra gold atom at the contact reduces the current significantly.Figure 2. Current-Voltage Characteristic

Experimental current-voltage characteristic of a benzene-1,4-dithiolate molecule as measured by Reed, et al. (top). Two theoretical curves corresponding to the benzene molecule attached directly to flat electrodes (middle) and to single gold atoms protruding from a surface (bottom). The extra gold atom at the contact reduces the current significantly.

APPLYING FULL QUANTUM MECHANICS

As a solid state physicist, Pantelides studies the behavior of atoms in solids, crystals, and surfaces, particularly how atoms rearrange themselves around defects in crystals or on surfaces. Defects cause atoms to move out of their "ideal" positions, and these rearrangements lead to the more interesting properties of materials. Such research has been at the forefront of computational materials physics for several decades. On the other hand, the calculation of the magnitude of electronic current through molecules and other nanostructures is an emerging frontier in computational physics.

In their studies of molecular nanodevices, Pantelides and Di Ventra have applied the full complexity of quantum mechanical methods, which required the development of new codes. Pantelides and Di Ventra have worked with Norton Lang at IBM's T.J. Watson Research Center to modify computer codes developed by Lang for use with scanning tunneling microscopes. In such a microscope, the magnitude of the current between the microscope's probe and the surface being studied describes the structure of the surface.

Pantelides and Di Ventra have modeled a two-terminal device constructed from benzene-1,4-dithiolate connected to gold or aluminum contacts at either end. Furthermore, they modeled different configurations of the contacts to determine how much of the current-voltage characteristic is due to the molecule and how much is due to the contacts.

"We've done things that are very difficult to replicate experimentally," Pantelides said. "For example, we've also looked at whether the molecule is attached to a flat surface or to one atom sticking out from the surface."

How to calculate the behavior of these systems was not well understood at the molecular level, particularly when a voltage is applied. Using a full quantum mechanical treatment has allowed them to see every significant behavior of the system. The computations have been conducted on NPACI's IBM SPs at SDSC and the University of Michigan. Computing a complete current-voltage curve for a single configuration requires about a thousand processor-hours.

"By using a full quantum mechanical treatment, we can not only compute the current for a particular voltage, but also calculate the forces on the atoms as the voltage increases," Pantelides said. "If the molecule is going to fall apart, we are going to see it."

Di Ventra, M., S.T. Pantelides, and N.D. Lang. 2000. First-principles calculation of transport properties of a molecular device. Physical Review Letters 84, 979–982.

Di Ventra, M., S.T. Pantelides, and N.D. Lang. 2000. The benzene molecule as a molecular resonant-tunneling transistor. Applied Physics Letters, 76, 3448–3450.

HANDLES FOR ENGINEERS

The results of the group's simulations have pointed to a sharp distinction in the behavior of the molecular devices, along with possible explanations for what causes the distinction. In particular, they discovered that the "signature" of the electronic properties of the molecule--the ups and downs of the current-voltage characteristic--are determined almost entirely by the molecule.

On the other hand, the magnitude of the current is determined by the contacts. When the molecule was connected to single gold atoms as the contacts, instead of a flat gold surface, the current passing through the molecule dropped by a factor of 100, while the signature remained the same. They then replaced the single gold atom with a single aluminum atom--and the current returned to its higher levels (Figure 2).

The difference results from a mismatch between the arrangements of electrons in the benzene and electrons in the gold and aluminum contacts. The outermost electron in a gold atom has a spherical distribution, while the outermost electron in an aluminum atom--and in the benzene molecule--has a dumbbell-shaped distribution. Therefore it is much easier to transport electrons between the aluminum and the benzene. On a flat gold surface (comprised of as few as three gold atoms) electrons have all the possible distributions so current is not restricted.

In molecules consisting of three benzene rings with an extra small molecule (called a ligand) attached to one of the rings, experiments by the Reed-Tour group showed a well-defined current-voltage spike. When measured at increasing temperatures, the spike kept moving to a lower voltage and broadening. The broadening at higher temperatures was expected; however, the shift to a lower voltage was a mystery. The Pantelides group's simulations showed that rotation of the ligands was the culprit. When the ligand molecules are rotated, an effect of higher temperatures, the shape change affects the energy levels of the structure and shifts the voltage at which the spike occurs. Seong-Gon Kim, another post-doctoral researcher who recently joined Pantelides' group, also participated in this project.

The regularity of the current-voltage signature demonstrates that molecules could be used as reliable electronic devices. The sharp spike in the three-molecule system presents even more interesting possibilities yet to be explored. And because they are predictable behaviors, the magnitude of the current due to the contacts and the shifting voltage spike in the three-molecule system give additional handles for designing useful electronic devices.

Finally, Pantelides and Di Ventra have begun simulating three-terminal devices built from molecular components. "We were able to add a third voltage, which distorts the distribution of electrons, and the current across the molecule was affected," Pantelides said. "Experimentalists haven't figured out how to apply such a third voltage, but we have demonstrated that you can get amplifications of current. So there is hope for creating such devices if you can figure out how to do this with real molecules." --DH *