Molecular Electronics

Spectacular science is coming out of research on molecular electronics.

In molecular-electronics research, molecules are used to yield the active and passive components (switches, sensors, diodes, resistors, LEDs, etc.) of electronic circuits or integrated circuits. For certain applications, such as molecular-based memory or logic circuitry, the devices are simply molecular-based analogues of devices with more conventional silicon-based circuitry. In those cases, molecular-electronics components may have the advantages of less manufacturing complexity, lower power consumption, and easier scaling. The field of molecular electronics is evolving rapidly, and even though there are no commercial applications as yet, the science coming out of this research is spectacular.

Consider the circuits of nanowires shown in the electron micrograph in Figure 1 (Melosh et al., 2003; see full [pdf] version for figures). The smallest (100-element) crossbar in this image is patterned at a density approaching 10¹²/cm², and the wire diameter is approximately 8 nm. With species like boron or arsenic, at a doping level of 10¹⁸/cm³, a similar 8-nm diameter micrometer-long segment of silicon wires would have 20 to 30 dopant atoms; a junction of two crossed wires would contain approximately 0.1 to 0.2 dopant atoms. Thus, conventional field-effect transistors fabricated at these wiring densities might exhibit nonstatistical, and perhaps unpredictable behavior. In fact, the patterning method (called superlattice nanowire pattern transfer [SNAP]) that produced the generation of patterns shown in Figure 1 can be used to prepare ultradense arrays of silicon nanowires. Thus, for the first time, researchers can interrogate the statistics of doping and other materials-type fluctuations that are expected to become important at the nanoscale.

Doping fluctuations, however, are relatively trivial compared to other problems encountered by engineers trying to scale conventional, silicon-based integrated circuits to significantly higher densities than are produced now. Power consumption (just from leakage currents through the gate oxide) is perhaps the most serious issue, but the lack of patterning techniques and high fabrication costs are also important.

In fact, no one is seriously contemplating scaling standard electronics device concepts to molecular dimensions (Packan, 1999); alternative strategies are being pursued, although they are all in early stages. These alternatives include molecular electronics, spintronics, quantum computing, and neural networks, all of which have so-called "killer applications." For quantum computing, it is the reduced scaling of various classes of NP-hard problems. For spintronics, it is a memory density that scales exponentially with numbers of coupled spin transistors. For molecular electronics, it is vastly improved energy efficiency per bit operation, as well as continued device scaling to true molecular dimensions. For true neural networks, it is greatly increased connectivity and, therefore, a greatly increased rate of information flow through a circuit.

Because molecular electronics borrows most heavily from current technologies, in the past few years it has advanced to the point that many major semiconductor-manufacturing companies, including IBM, Hewlett-Packard, and LG (Korea), are launching their own research programs in molecular electronics.

At device areas of a few tens of square nanometers, molecules have a certain fundamental attractiveness because of their size, because they represent the ultimate in terms of atomic control over physical properties, and because of the diverse properties (e.g., switching, dynamic organization, and recognition) that can be achieved through such control.

In the crossed-wire circuit shown in Figure 1 (called a crossbar circuit), the molecular component is typically sandwiched between the intersection of two crossing wires. Molecular-electronics circuits based on crossbar architectures can be used for logic, sensing, signal routing, and memory applications (Luo et al., 2002). To realize such applications, many things must be considered simultaneously: the design of the molecule; the molecule/electrode interface; electronically configurable and defect-tolerant circuit architectures; methods of bridging the nanometer-scale densities to the submicrometer densities achievable with lithography; and others (Heath and Ratner, 2003). Using a systems approach in which all of these issues are dealt with consistently and simultaneously, we have been able to fabricate and demonstrate simple molecular-electronics-based logic, memory, and sensing circuitry.

The active device elements in these circuits are molecular-mechanical complexes (Figure 2) organized at each junction within the crossbar, as shown in Figure 3 (Heath and Ratner, 2003; Luo et al., 2002). The molecular structure in Figure 2 (left) is a [2]catenane that consists of two mechanically interlocked rings. One ring is a tetracationic cyclophane (TCP⁴⁺); the other is a crown-ether-type ring with two chemical recognition sites, a dioxynapthyl (DN) group and a tetrathiafulvalene group (TTF). The structure on the right side of Figure 2 is a [2]rotaxane that consists of similar chemical motifs. Here the TCP⁴⁺ ring encircles a dumbbell-shaped structure that has both DN and TTF recognition groups. The molecules are switched via a one- or two-electron process that results in a molecular-mechanical transformation (and a significant change in the electronic structure) of the molecule. This type of molecular actuation, which can be rationally optimized through molecular design and synthesis, provides the basis for information storage or for defining the "open" and "closed" states of a switch.

Typical data from one of our molecular switches is shown in Figure 4 (we have incorporated additional data from various control molecules in this figure). For the structures shown in Figure 2, the controls include [2]catenanes with identical recognition sites (i.e., two DN groups), the dumbbell component of the [2]rotaxane structure, the TCP⁴⁺ ring, and others. One always does control experiments, of course, but controls are critical here, because these devices are difficult to characterize fully, and one of the few experimental variables is molecular structure. Perhaps the most fundamental challenge facing scientists constructing solid-state molecular-electronic devices and circuits is developing an intuition for guiding the design of the molecular components.

Charge transport through molecules has been known and studied for a long time, but it has traditionally been a solution-phase science (Joachim et al., 2000; Kwok and Ellenbogen, 2002; Mujica and Ratner, 2002; Nitzan, 2001; Ratner, 2002). For example, a critical component of electron-transfer theory in molecules is the solvent-reorganization coordinate. Molecular synthesis is also a solution-phase endeavor, and all of the analytical techniques (e.g., mass spectrometry, NMR, optical spectroscopy) are primarily designed to investigate the structure and dynamics of molecules in solution. When molecules are sandwiched between two electrodes, none of these theories and analytical techniques is of any use, and new concepts must be developed.

This situation is exacerbated because certain critical aspects of basic solid-state-device physics cannot be translated directly into the world of molecular electronics. For example, consider the following two fundamental tenets of solid-state materials. First, when two different materials are brought together, their Fermi levels align with each other. Second, if a conducting wire of length L has a measured resistance of R, then a wire of the same material and diameter, but with a length 2L, will exhibit a resistance of 2R. This is called ohmic conductance. Neither of these rules holds true for molecules. This is because molecular orbitals are spatially localized, whereas in solids the atomic orbitals form extended energy bands. When a molecule is adsorbed or otherwise attached to the surface of a solid, there is no straightforward way to think about the position of the molecular orbitals with respect to the Fermi level of the solid. Furthermore, if a molecule of length L yields a resistance value of R, then a similarly structured, but longer molecule could actually be a better conductor or, perhaps, a far worse conductor. There is no real ohmic conductance in molecules.

Thus, the fundamental challenge is to develop an intuition of how molecules behave in solid-state settings and to use that intuition as feedback to molecular synthesis. In fact, we need a method of characterizing molecular-electronic devices that yields the type of rich information generated by modern NMR spectroscopic methods. Perhaps the most promising approach is single-molecule, three-terminal devices in which a single molecule bridges a very narrow gap (called a break junction) between a source and a drain electrode. A third electrode (called the gate) provides an electric field for tuning the molecular-electronic energy levels into and out of resonance with the Fermi energies of the source and drain electrodes. These devices were originally developed by Park and McEuen and were then further explored by their two groups separately (Liang et al., 2002; Park et al., 2002). Figure 5 shows data from a single-molecule device containing the dumbbell component of the [2]rotaxane molecule shown in Figure 2 (Yu et al., 2003). These data show clearly that these types of device measurements represent a very high information-content analytical method. In an analogy to optical methods, the energy levels resolved in such a device span a very broad range of the spectrum, from ultraviolet to far infrared. This means that molecular orbital energies, and even low-frequency molecular vibrations, might be observable. How molecular orbitals align with the Fermi levels of electrodes, the coupling of molecular vibrations with charge transport through the molecule, and the nature of the molecule/electrode interface states are questions being addressed in various laboratories around the world.

Some of the critical length scales of circuits that can now be fabricated, such as the diameter of the wires and the interwire separation distance (or pitch), are more commonly associated with biological macromolecules, such as proteins, mRNA oligonucleotides, and so on, than with electronics circuitry. In fact, one unique application of nanoscale molecule-electronics circuitry, and perhaps the application that sets this field apart from traditional electronics, is the potential construction of an electrical interface to a single biological cell (Cui et al., 2001). One of our ongoing projects is constructing such an interface designed to measure, simultaneously and in real time, thousands of molecular signatures of gene and protein expression (Heath et al., 2003; Zandonella, 2003). This type of circuitry may eventually provide a direct connection between the worlds of molecular biology and medicine and the worlds of electrical engineering and integrated circuitry.

Acknowledgments
The work described in this paper has been supported by the Defense Advanced Research Projects Agency, the MARCO Center, the Semiconductor Research Corporation, the Office of Naval Research, the U.S. Department of Energy, and the National Science Foundation. The results presented here were collected by an outstanding group of students and postdoctoral fellows at UCLA and Caltech.

References
Collier, C.P., G. Mattersteig, E.W. Wong, Y. Luo, K. Beverly, J. Sampaio, F.M. Raymo, J.F. Stoddart, and J.R. Heath. 2002. A [2]-catenane based solid-state electronically reconfigurable switch. Science 289(5482): 1172-1175.

Cui, Y., Q. Wei, H. Park, and C.M. Lieber. 2001. Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species. Science 293(5533): 1289-1292.

Heath, J.R., and M.A. Ratner. 2003. Molecular electronics. Physics Today 56(5): 43-49.

Heath, J.R., M.E. Phelps, and L. Hood. 2003. Nanosystems biology. Molecular Imaging and Biology (accepted and in press, 10/03).

Joachim, C., J.K. Gimzewski, and A. Aviram. 2000. Electronics using hybrid-molecular and mono-molecular devices. Nature 408(6812): 541-548.

Kwok, K.S., and J.C. Ellenbogen. 2002. Moletronics: future electronics. Materials Today 5(2): 28-37.

Liang, W., M.P. Shores, M. Bockrath, J.R. Long, and H. Park. 2002. Kondo resonance in a single-molecular transistor. Nature 417(6890): 725-729.

Luo, Y., C.P. Collier, K. Nielsen, J. Jeppesen, J. Perkins, E. DeIonno, A. Pease, J.F. Stoddart, and J.R. Heath. 2002. Two-dimensional molecular electronics circuits. ChemPhysChem 3(6): 519-525.

Melosh, N.A., A. Boukai, F. Diana, B. Geradot, A. Badolato, P.M. Petroff, and J.R. Heath. 2003. Ultrahigh-density nanowire lattices and circuits. Science 300(5616): 112-115. Published online March 13, 2003 (10.1126/science.1081940).

Mujica, V., and M.A. Ratner. 2002. Molecular Conductance Junctions: A Theory and Modeling Progress Report. Pp. 10-1-10-27 in Handbook of Nanoscience, Engineering, and Technology, D. Brenner, S. Lyshevski, G. Iafrate, W.A. Goddard III, eds. Boca Raton, Fla.: CRC Press.

Nitzan, A. 2001. Electron transmission through molecules and molecular interfaces. Annual Review of Physical Chemistry 2(1): 681-750.

Packan, P. 1999. Pushing the limits. Science 24(5436): 2079-2081.

Park, J., A.N. Pasupathy, J.I. Goldsmith, C. Chang, Y. Yaish, J.R. Petta, M. Rinkoski, J.P. Sethna, H.D. Abru?a, P.L. McEuen, and D.C. Ralph. 2002. Coulomb blockade and the Kondo effect in single-atom transistors. Nature 417(6890): 722-725.

Ratner, M.A. 2002. Introducing molecular electronics. Materials Today 5(2): 20-27.

Yu, H., Y. Luo, H.R. Tseng, K. Beverly, J.F. Stoddart, and J.R. Heath. 2003. The molecule-electrode interface in single-molecule transistors. Angewandte Chemie (accepted and in press, 10/03).

Zandonella, C. 2003. Cell nanotechnology: the tiny toolkit. Nature 423(6935): 10-12.

James R. Heath is the Elizabeth W. Gilloon Professor of Chemistry at the California Institute of Technology.