September 30, 2002
Volume 80, Number 39
CENEAR 80 39 pp. 38-43
ISSN 0009-2347


NANOSCALE ELECTRONICS
Bustling research is producing sophisticated laboratory demonstrations, but commercialization of nanometer-sized devices remains a ways off

MITCH JACOBY, C&EN CHICAGO

At first there were only a few of them, but recently, their numbers have multiplied wildly. Newspaper headlines, magazine articles, journal papers, even television commercials now are loaded with those big "nano" words: nanometer, nanoscale, nanosecond, and nanotechnology, to name a few. And it seems that every week some organization is announcing yet another "nanoconference."

Why all the excitement? Because in recent years, as scientists have begun to get a handle on controlling matter at the nanometer scale, they've recognized that these skills can be used to make new materials with unique and useful properties--perhaps leading to a range of commercial applications in sensing, electronics, and other areas. Impressive laboratory demonstrations of nanoscale dexterity have been widely reported in the scientific and popular press, thereby drawing the attention of more scientists, companies, and technology investors.

Nanoelectronics--a developing field in which circuitry is composed of nanometer-sized electronic components--is one topic that's attracting a lot of interest in scientific and nonscientific corners alike. Researchers are driven to explore the field because further miniaturizing today's already small electronic circuits will lead to faster, more sophisticated, and more portable devices. Yet it's widely believed that in a decade or so, silicon-based circuitry will have been shrunk as small as is physically possible. And so the search is on for alternative materials from which nanoscale circuits can be constructed.

VIDEO - CLICK ON IMAGE
Switching of a catenane Switching of a rotaxane in a device setting Construction of a crossbar device

Carbon nanotubes and nanowires of materials such as gallium arsenide are being studied for just that purpose by several research groups. Other investigators, whose work is often labeled "molecular electronics," focus on the electrical behavior of individual molecules or small numbers of molecules. Scientists who work in these fields aim to design and build new types of computers in which the nanosized entities lie at the heart of logic circuits. Skeptics say that the proposed applications are exaggerated and that reports are full of "hype." But researchers around the globe are exploring and discovering nanometer-scale phenomena, and they are rapidly turning out large numbers of scientific papers.

"These are exciting and heady times," says Mark A. Ratner, surveying the combined topics of molecular and nanoelectronics. "There are wonderful experiments being reported these days," he says. "The excitement is justified." A nearly 30-year-old paper by the Northwestern University chemistry professor is often cited as the beginning of modern molecular electronics. In that theoretical study, Ratner and Arieh Aviram, who was at that time a graduate student, proposed that under certain conditions, individual organic molecules could function as p-n junction diodes. These simple electronic devices typically are formed at the interface between positive charge-carrying (p-type) and negative charge-carrying (n-type) semiconductors and can be used as rectifiers to convert AC current to DC current.

Ratner notes that few advances were made in molecular electronics until scanning probe microscopes became available in the late 1980s. The new tools gave the field a much-needed boost by making it possible to measure a whole range of phenomena, including the type Ratner had described in the 1970s. He adds that progress in other areas, for example, in methods for preparing self-assembled monolayers of molecules, was also instrumental in jump-starting nanoscale electronics. "Now the advances are coming thick and fast," he says.

"Thick and fast" aptly describes the flurry of research being carried out on carbon nanotubes. According to Phaedon Avouris, manager of nanoscale science and technology at IBM's T. J. Watson Research Center, Yorktown Heights, N.Y., the strawlike all-carbon structures with nanometer-sized diameters are endowed with unique properties that make them excellent candidates for nanoelectronic applications.

 
FULL SIZE - CLICK IMAGE
TINY WORLD Molecules that can be switched between conducting and nonconducting states--such as the UCLA-synthesized [2]rotaxanes shown here--lie at the heart of some of today's lab demonstration circuits. In one state (right structure and upper right schematic), a cationic cyclophane ring (blue) encircles a tetrathiafulvalene (TTF) unit (green). Oxidizing the TTF unit causes electrostatic forces to drive the cyclophane ring to a dioxynaphthalene unit (red), switching the molecule to the other state (left).
COURTESY OF UCLA

NANOTUBES CAN BE metals or semiconductors, depending on their chirality, Avouris notes. And because of their strong chemical bonds and satisfied valencies, the materials boast high thermal, mechanical, and chemical stability. In addition, carbon nanotubes can be efficient conductors as a result of their tiny diameters, long lengths, and defect-free structures that make them ideal one-dimensional systems.

Just four years ago, Avouris' research group and, independently, scientists working with Cees Dekker, a professor of applied physics at Delft University of Technology, in the Netherlands, demonstrated that field-effect transistors (FETs) could be fashioned from carbon nanotubes. FETs are a type of switch in which a semiconducting channel bridges two electrodes designated "source" and "drain." Current flow between these electrodes is controlled by a third electrode known as a "gate." By applying a voltage to the gate, the semiconductor's state can be changed--reversibly--from insulating to conducting, thereby switching the transistor on or off. In conventional FETs, the bridging channel is made of silicon. In nanotube devices, the channel is a single carbon nanotube.

Today's sophisticated silicon-based integrated circuits owe much of their success to advances in FET technology. Manufacturers pack tens of millions of the tiny transistors into the postage-stamp-sized computer chips used in modern microprocessors. For that reason, the IBM research group and others have been working on improving the properties of nanotube FETs. Much of the work focuses on understanding and controlling electrical conduction through the tiny tubes.

One problem that plagues researchers looking to fashion circuit components from nanotubes is separating metallic tubes from the ones that are semiconducting. Common synthesis procedures produce spaghetti-like mixtures of nanotube ropes that are unusable for semiconductor applications because they contain both types of tubes.

But last year, Avouris and coworkers came up with a way to sort through the tangle. After fashioning electrodes around nanotube bundles via lithography methods, the team applied certain voltages to the gate electrodes to switch off the semiconducting tubes, converting them to insulators. Then, by applying high voltage to the circuit, the IBM group oxidized the metallic tubes, causing them to break down without affecting the semiconducting tubes. The group used the method to prepare FET arrays from single-walled nanotube ropes and to peel apart multiwalled nanotubes shell-by-shell (C&EN, April 30, 2001, page 13).

Just a few months later, the IBM group showed that elementary computing circuits known as logic gates--which typically are constructed from combinations of FETs--can be fashioned from a single nanotube bundle. Shortly thereafter, Delft's Dekker reported similar advances in constructing multi-FET nanotube logic circuits. Complex combinations of AND, OR, and NOT logic gates make up the core of digital processors.

To make NOT gates using nanotubes, Avouris and coworkers needed to come up with a procedure for preparing n-type nanotubes. NOT gates require both types of transistors. But the synthesis techniques available at that time, produced only p-type tubes. Hunting for a solution, Avouris and IBM coworkers Vincent Derycke, Richard Martel, and Joerg Appenzeller found that p-type nanotubes can be converted to n-type simply by annealing (heating) in vacuum. Armed with the new preparation trick, the team prepared separate n- and p-type transistors and fabricated a NOT gate.

Then they went a step further and fabricated a NOT gate using just one nanotube. For that feat, the team selectively converted a short length of a nanotube to n-type by doping it with potassium through a tiny window in a protective polymer coating while leaving the unexposed portion p-type. The single tube with p- and n-type segments was then used to construct an intramolecular NOT gate (C&EN, Sept. 3, 2001, page 9).

8039covstory
Ratner
8039covstory
Avouris
PHOTOS BY MITCH JACOBY
FURTHER ADVANCES have come from transistor design modifications. Earlier this year, for example, the IBM group reported that their so-called top-gated nanotube FETs outperform state-of-the-art silicon FETs in terms of switching rate and the amount of current they can carry per width of conductor. One key difference between the latest design and earlier designs is that the silicon wafers that support the FETs do not function as gates. Instead, the gate is fabricated above the nanotube--allowing all FETs in contact with the silicon wafer to be switched independently. In addition, the new FET design benefits from switching voltages that are an order of magnitude lower than those needed to switch older FETs [Appl. Phys. Lett., 80, 3817, (2002)].

Just recently, the IBM team developed a catalyst-free procedure for preparing single-walled carbon nanotubes. Conventional methods for preparing single-walled tubes, such as discharge and vapor deposition techniques, rely on particles of nickel and cobalt or other catalytic transition metals. A problem with those procedures is that the metal particles left behind in the products disturb the nanotubes' electrical properties, forcing scientists to take complex purification measures.

But now Avouris and coworkers have shown that single-walled nanotubes can be prepared from SiC in the absence of metal catalysts at temperatures above 1,500 °C. The method produces nanotubes that are 1.2 to 1.6 nm wide and are aligned along the face of the support. Other researchers have reported techniques for growing vertically aligned products, which they dub nanotube forests. But nanotubes that grow horizontally are amenable to standard vapor deposition methods and may be connected in parallel to lower their electrical resistance [Nano Lett., published online, Sept. 24, http://dx.doi.org/10.1021.nl0256309].

Despite the recent advances, Avouris asserts that nanotube FETs remain "far from optimized." Improvements could be made by using thinner insulators with higher dielectric constants, he suggests. "But what's really needed is a better understanding of the mechanism of electrical switching in nanotube FETs."

Carbon nanotubes aren't the only game in town. Nanowires made of semiconductors such as silicon, gallium arsenide, and indium phosphide are being investigated as candidate materials for nanoelectronics. The field is advancing rapidly as researchers are making fast progress in synthesis, device fabrication, and testing.


Skeptics say that the proposed applications are exaggerated and that reports are full of "hype."

Lieber
Lieber
COURTESY OF CHRISTOPHER NAVIN
Although Harvard University chemistry professor Charles M. Lieber has published widely on carbon nanotubes, semiconducting nanowires figure prominently into his research program. The tiny wires are versatile building blocks that can be used in a bottom-up approach to constructing nanoelectronic circuits, Lieber notes, because the size, structure, and functional properties of nanowires can be controlled readily via synthesis procedures.

In 1998, Lieber and coworkers described a vapor-liquid-solid synthesis method in which laser light is used to ablate nanometer-sized metal clusters that serve as nucleation centers and catalysts for nanowire growth. The Harvard researchers used the technique to prepare uniform single-crystalline nanowires of silicon and germanium with diameters as small as 3 nm and lengths up to 30 mm. Since that time, the procedure has been used to prepare nanowires with even smaller diameters, and it's been extended to a wide variety of materials. Examples include III-V semiconductors such as GaAs and II-VI semiconductors such as ZnSe.

Recently, a number of research groups boosted the complexity of materials that can be prepared by the cluster-nucleation method. The teams prepared modulated structures--nanowires composed of dissimilar segments. The two-tone materials are made by turning the supplies of reactants on and off during synthesis with pulsed lasers or by other methods. These "heterostructured" products open the door to new sophisticated applications, such as terahertz-frequency photon emitters.

Lieber, graduate student Mark S. Gudiksen, and coworkers used modulation methods to prepare nanowires with 21 alternating segments of GaAs and GaP. And they prepared Si and InP nanowires with modulated doping, such that the wires were endowed with alternating p- and n-type regions. Using similar methods, Peidong Yang, an assistant chemistry professor at the University of California, Berkeley, prepared Si-SiGe nanowires. And Lars Samuelson of Lund University, in Sweden, synthesized InAs-InP nanowires (C&EN, Feb. 11, page 7).

 
Nanotubes_Avouris
STOP METALING A catalyst-free synthesis procedure developed at IBM produces aligned single-walled carbon nanotubes (vertical lines) from SiC without residual metal impurities.
COURTESY OF IBM RESEARCH

TO EVALUATE the usefulness of the tiny wires in nanoscale electronics, the Harvard researchers fashioned the wires into test circuits that were fabricated using lithography methods and then studied the circuits' electrical properties. According to Lieber, Si nanowire FETs, for example, perform remarkably well. The devices exhibit large on/off ratios (>105)--a requirement for effective switches--and charge-carrier mobilities that exceed those measured in conventional silicon devices.

In addition to constructing FETs from nanowires, Lieber's team has built and tested photodetectors, biological and chemical sensors, and more complex devices, including light-emitting diodes (LEDs) and logic gates. Most of the papers in which the results are reported were published in just the past 18 months.

In the chemical sensing study, Lieber, graduate students Yi Cui and Qingqiao Wei, and Harvard assistant chemistry professor Hongkun Park found that by functionalizing boron-doped Si nanowires, they could prepare efficient sensors. Modifying the wires with an amino-silane compound, for example, produced pH sensors that are linear over a large dynamic range. And they observed that biotin-modified wires can be used to detect picomolar levels of streptavidin. In addition, the Harvard chemists found that the wires could be used to detect proteins reversibly in real time [Science, 293, 1289 (2001)].

By crossing an n-type and a p-type InP nanowire and then attaching electrodes to the wires lithographically, Lieber and coworkers made a nanometer-sized junction--at the point of overlap--that behaved as a current rectifier and an LED [Nature, 409, 66 (2001)]. Based on the study, Lieber says that more complex devices can be constructed from semiconducting nanowires simply by crossing them and forming tic-tac-toe patterns. But taking advantage of the crossed-nanowire idea requires developing a method for preparing the patterns efficiently and reproducibly.

By combining surface patterning techniques and microfluidic alignment methods, the Harvard group developed the needed procedure. Their idea is to flow nanowires from a suspension into tiny parallel channels in a mold, in a layer-by-layer fashion, while rotating the mold orientation--and hence, nanowire flow direction--between layers [Science, 291, 630 (2001)].

With the crossed-nanowire method for making p-n junctions firmly in hand, Lieber and coworkers were well on their way toward making computational logic circuits. Near the end of 2001, the group published a study in Science showing that their procedures could be used to make arrays of nanowire FETs configured as AND, OR, and NOR logic gates. Lieber stresses that in crossed-nanowire structures, such as the Harvard logic gates, the width and length of the FET channels are controlled by synthesis and assembly--not lithography--unlike today's microelectronic components.

 


Only a few years ago, many scientists sat through molecular electronics presentations thinking that the far-fetched nanocomputing goals would never be reached.

TEAM EFFORT Molecular electronics research at UCLA brings together organic synthesis and device measurements through the efforts of (standing, from left) Jeppesen, Stoddart, Heath, and Yi Luo, and (kneeling, from left) Tseng, Kristen Beverly, and Paul Celestre.
COURTESY OF UCLA
THESE FABRICATION concepts make it possible "to envision a straightforward path to nanometer-scale electronics of the future," Lieber says. His vision, which is supported by experiments conducted by his group in the past several months, includes a nanoscale logic device known as a decoder that can address individual elements in crossed-wire arrays.

Using a handful of molecules--or even just one of them--to make nanoscale switches or other devices is an idea that's gathered momentum in recent times. In 1999, researchers working with chemistry professors James R. Heath and J. Fraser Stoddart of the University of California, Los Angeles, demonstrated that simple circuits--fuses--could be made from a layer of [2]rotaxane molecules. Then a team headed by Rice University chemistry professor James M. Tour and Yale University electrical engineer Mark A. Reed reported success in making reversible switches using benzene thiol molecules.

Other groups have also been investigating the possibilities of single-molecule systems. For example, a team at UC Berkeley including Park (now at Harvard), Paul L. McEuen (now at Cornell), and A. Paul Alivisatos, trapped a single C60 molecule in a tiny gap between a pair of electrodes to study electronic conduction mechanisms [Nature, 407, 57 (2000)]. And just recently, a single molecule of a transition metal-organic complex--containing either one Co atom or two V atoms--was turned into a single-molecule transistor in independent studies by Park and McEuen (C&EN, June 17, page 4).

Before molecular devices can be wired up and tested, organic chemists need to synthesize the molecules that drive them. In the UCLA collaboration, Stoddart's group prepares the compounds and Heath's group turns them into circuit components. To make the molecules switchable, Stoddart's group designs them so that one segment of a molecule can move (rotate or translate) relative to another segment.

Stoddart notes that, contrary to remarks made by many researchers, the molecules are not difficult to prepare nowadays. But it took some years to build up the necessary expertise in molecular recognition concepts, p-donor/p-acceptor interactions, and other subjects. He adds that his group has synthesized "hundreds, if not thousands" of catenane and rotaxane analogs since 1989.

In 2000, the UCLA team demonstrated a reusable switch based on a [2]catenane--a molecule with a pair of interlocked rings. The switching state of the redox-active molecule was shown to change reversibly, depending on an applied potential. Electrostatic repulsion drives rotation of one ring relative to the other between open and closed states.

 
FULL SIZE - CLICK IMAGE
MAGNITUDE A series of micrographs in which each image is magnified roughly 10 times more than the previous one (left to right) reveals details of a new HP Labs molecule-based 64-bit prototype memory and logic chip. Using a newly patented procedure, researchers patterned a silicon wafer with 625 memory devices (one shown at top right). At the center of each memory circuit, an 8 3 8 array of crossed metal nanowire electrodes (bottom row) sandwiches roughly 1,000 rotaxane molecules forming 64 bits of nonvolatile memory.
COURTESY OF HEWLETT-PACKARD LABS

LOOKING TO IMPROVE device performance by tailoring the molecules, Jan O. Jeppesen, a visiting graduate student in Stoddart's lab, and postdoctoral associate Hsian-Rong Tseng prepared various rotaxane analogs of the catenane that brought success in 2000. The series of amphiphilic bistable [2]rotaxanes are dumbbell-shaped molecules capped with large stoppers that are threaded through a cyclophane ring. As with the [2]catenanes, the ring's position relative to the other piece of the molecule is controlled by redox interactions.

Heath and coworkers constructed circuits with the new molecules using a layer-by-layer deposition method to fabricate tic-tac-toe structures with a few thousand molecules situated at each intersection. The team used this fabrication method to prepare 64-bit random-access memory circuits [ChemPhysChem, 3, 519 (2002)]. And Hewlett-Packard Labs announced earlier this month that R. Stanley Williams, Yong Chen, and coworkers have also succeeded in using the UCLA [2]rotaxane molecules to fabricate a 64-bit memory circuit.

Only a few years ago, many scientists sat through molecular electronics presentations thinking that the far-fetched nanocomputing goals would never be reached. But the flurry of demonstration results in just the past two years appears to have decreased the number of nonbelievers. Yet some naysayers remain. Just recently, a California professor was furious about a C&EN story on molecular electronics and wrote to the editor, exclaiming, "In reality, there is no such topic!"

The professor has a few surprises in store. In reality, molecular and nanoelectronics are bustling fields. Large numbers of commercial products may not be available in the near future, but as the vibrant research activity reveals, they're coming.

FULL SIZE - CLICK IMAGE
8039cov1
MODULAR MOLECULES UCLA chemists use synthesis techniques to prepare multipart molecules with customized structures and properties for nanoelectronic devices. [2]Catenanes (left) and [2]rotaxanes (center and right) can be switched on and off depending upon redox state, which determines where in the molecule a cyclophane ring (blue) resides.
COURTESY OF UCLA
MORE ONLINE
Links to videos of the following processes can be found on C&EN Online, http://pubs.acs.org/cen/video/8039/8039nanovideo.html.

* Switching of a catenane

* Switching of a rotaxane in a device setting

* Construction of a crossbar device



Top


Chemical & Engineering News
Copyright © 2002 American Chemical Society