Getting Nanowired.

Makers of nanowires may overcome the limits that loom for microchip fabrication 

In his laboratory at Harvard University, chemist Charles M. Lieber puts a drop of clear solution onto a tiny silicon wafer. It looks like any old liquid, but it isn't. Suspended in the fluid are millions of wires, tens of thousands of times thinner than a human hair. These wires could become the building blocks of smaller, cheaper, and faster electronics.

Using these nanowires, Lieber and other physicists, chemists, and engineers are trying to break the barrier that faces today's methods of fabricating the fine circuit components that go into computer chips. These current methods can't shrink the components to dimensions measured in billionths of a meter, which are required to make chips for more powerful computers.

With nanowires, however, researchers have made prototypes of some of the world's tiniest electronic devices, such as diodes and transistors. Transistors amplify electronic signals, and diodes keep electrons flowing in one direction. Now, researchers are tackling perhaps the biggest challenge: integrating these minuscule devices into a network of circuits that could power a new generation of computers.

The heart of a computer is a fingernail-sized chip, or integrated circuit, that includes transistors, diodes, and other devices that can manipulate electrons. To keep up with the demand for faster and more powerful computers, engineers have had to squeeze more circuitry onto the chip.

In conventional chip fabrication, scientists slice a block of silicon into thin wafers. Then, with a series of steps that modifies the silicon by etching it away or depositing materials on top of it, they create tiny, complex circuitry. Under a microscope, the result can look like a dense metropolis. Intel's newest chip, the Pentium 4, holds roughly 42 million transistors, with each transistor spanning roughly 130 nanometers (nm).

That's not an easy or inexpensive accomplishment. Each additional level of miniaturization has added to the chip makers' expenses. It costs at least $2 billion to build a chip-manufacturing plant today, says Philip J. Kuekes, a computer architect at the Hewlett Packard Laboratory in Palo Alto, Calif. In 10 years, that cost will be about $50 billion, Kuekes says. Moreover, at a certain point of miniaturization, quantum laws take over and prevent devices from working according to the classical laws of physics that govern the operation of conventional chips.

That's where nanostructures come in, some researchers say. "Instead of using this very expensive process of taking a big chunk of material and cutting it up and throwing most of it away to make the structure, we thought, Why don't we make these very small pieces and then organize them" into devices? says Lieber.

The small pieces can be made several ways. Some researchers are making tiny tubes, called bucky tubes, from carbon (SN: 5/9/98, p. 294). Others are making nanowires from the same materials that revolutionized microelectronics: semiconductors and metals.

By using a range of semiconducting materials such as silicon, germanium, zinc oxide, indium phosphide, and molybdenum selenide, Lieber, Peidong Yang of the University of California, Berkeley, and other researchers are making wires only tens of nanometers wide and micrometers long. Yang is also making nanowires out of metals such as gold, silver, platinum, and copper.

"We must explore all these different nanowires to put together a nano-toolbox with which we can assemble different devices on the same chip," notes Yang.

While the semiconductor nanowires will be useful in building devices, such as transistors, the metal wires will find their niche when it comes time to wire the devices into an integrated circuit. "There's no sense in making very small nanoscale components if you cannot wire them together," says Uzi Landman, a physicist at the Georgia Institute of Technology in Atlanta.

Researchers at Pennsylvania State University in State College are making "striped" nanowires with a semiconducting segment sandwiched between metallic segments. This enables chemist Thomas E. Mallouk and his coworkers to build specific types of electronic devices directly within the nanowires. For example, the researchers expect these more complex nanowires to preserve the information traveling between components better than simpler nanowires do.

Striped wires are actually quite easy to create, says Mallouk, whose freshman chemistry students have learned to do it.

The first step is to grow metallic nanowire segments in an electroplating process. Mallouk and his coworkers use a special filter made out of alumina, a tough ceramic material. First, they coat one side of the filter with silver. Next, they put the filter, silver-side down, into a container and pour a solution containing ions of another metal, such as gold or copper, atop the uncoated side. When the researchers apply a voltage to the filter, the dissolved ions are driven through the filter's tiny pores and atoms begin to form into wires, says Mallouk.

To create the semiconducting layer, the researchers add a polymer, which sticks to the tips of the metal segments. Then, the chemists apply a solution of semiconducting particles, such as cadmium selenide. These particles migrate to the polymer and pile up to form the semiconducting segment of the nanowire. Another electroplating step completes the job by adding the final metallic segment.

With these techniques, Mallouk and his coworkers have grown wires that are 1 to 10 micrometers long and as thin as 15 nm across.

To grow even thinner wires, some researchers are incorporating only semiconducting materials, such as silicon. Lieber produces his wires by using tiny clusters of metal atoms. He melts these clusters in a hot reaction chamber and then adds atoms of semiconductor, which mix with the metal into a larger droplet. "If you keep adding silicon, the metal ultimately becomes super saturated, in the same way you can keep adding salt to water," Lieber explains.

When that happens, pure silicon atoms build up and extend from the droplet like whiskers, says Lieber. With this technique, he and his coworkers have created wires as narrow as 2 nm.

Moreover, by changing the materials he uses, Lieber can make different devices such as transistors and diodes. He and his colleagues described this work several years ago.

Under many conditions, pure semiconductors do not conduct electricity, says Lieber. To fix that, he and his coworkers must dope the material with other atoms that establish pathways for current, he explains. For example, they add boron atoms to make the material p-doped, which means the material can accept electrons. Adding phosphorus leads to n-doped materials, which can give away electrons. When the researchers bring p-doped and n-doped materials together, electrons can flow.

After Lieber and his coworkers create their doped wires, they place them in a solution to manipulate them. Says Lieber, "We can set up on our lab bench little bottles with solutions of different building blocks, and we can take these building blocks and build up devices from a surface," such as a silicon wafer.

In the Feb. 2 SCIENCE, Lieber's group reported using its silicon nanowires to create a bipolar transistor, a workhorse of microelectronics technologies.

To make the transistor, Lieber crosses three silicon wires. He sets two n-doped wires side-by-side and then lays a p-doped wire across them. The points where the wires cross can regulate the flow of electric current.

A month earlier, in the Jan. 4 NATURE, Lieber and his coworkers reported using indium phosphide nanowires to create come of the world's tiniest diodes. They emit light just as the semiconductor diodes of everyday LEDs, or light-emitting diodes, do.

The new nanoscale LEDs, however, exhibit novel behaviors that could make them more versatile than the conventional LEDs found in displays on electronic devices like cell phones. For example, Lieber hopes that by varying the thickness and composition of his nanowires, he can make LEDs that give off any color. When arranged into a matrix, nanoscale LEDs could have optoelectronic applications in areas ranging from telecommunications to information storage.

"I think it's a very, very nice step forward to show that you actually can integrate [the wires] into a device," says Landman. However, he cautions, engineers might think of each nanowire structure as "a physicist's toy" rather than a device. It will take engineering to turn such toys into practical technologies, he says.

Some researchers are taking steps toward that goal. For example, Yang and his colleagues are using their nanowires to make nanolasers. These light-beaming devices could speed data processing and improve lab-on-a-chip devices that might help biotechnology researchers analyze DNA samples more efficiently, says Yang.

While making these electronic nanodevices is tough, an even more daunting task is to put them together into a circuit, contends Mark Reed of Yale University. "If you're going to sell this in your favorite electronics store ... it has to be integrated into a system," he says.

In today's computer chip, each device is positioned in a specific spot on the silicon wafer. This precision, however, would be impractical in the nanoworld, says Lieber. On this small scale, he says, even tiny temperature-driven jostlings can lead to defects. Some defects can be tolerated, but when too many accumulate, the circuitry will fail.

One way to reduce the number of defects is to simplify the architecture of the circuits, says Lieber. Then, electrons will be able to find appropriate pathways around flaws. An additional advantage of this approach would be a great reduction in the power consumption as the chip operates, he adds.

To build their nanowires into rudimentary circuits, Lieber and his coworkers are using fluids. They start by putting a drop of solution containing millions of p-doped nanowires onto a silicon wafer. Tiny tunnels scooped into what Lieber likens to a rubber stamp then guide the fluid so that all the wires end up side-by-side, like a layer of toothpicks in a box.

After the first layer of wires dries, the researchers turn the stamp 90 [degrees] degrees and put a drop containing n-doped nanowires onto the surface. Lieber says that nearly all the wires of the top layer cross over those of the bottom layer, each junction forming a potential diode.

Using this technique, the researchers can make networks of simple devices. They plan to wire these devices together to the create complex structures that perform memory and logic operations.

Flowing wires is just one strategy for organizing nanodevices. For example, Mallouk and his colleagues are using DNA to get their wires where they want them.

"DNA in solution is incredibly selective," says chemist Christine D. Keating, who works with Mallouk. "One strand can find its matching strand amidst a sea of thousands of others that don't match." When bound to certain metals in nanowires, the DNA acts as a kind of "smart glue," she explains.

One problem the researchers have encountered is that DNA on a wire fails to recognize its matching strand about 20 percent of the time. To build real technologies using DNA in this way, Keating says, the researchers will need to improve their "hit rate."

Mallouk and his coworkers have already used this technique to build nanowires into T-shape structures, crosses, and triangles, which may someday go into such complex devices as memory components.

With the work on nanowires making impressive progress, Lieber anticipates some near-term applications. "If one could have some very sensitive sensor which could be continuously monitoring for specific compounds in the blood, then one would have a huge opportunity in health care," says Lieber.

For example, patients with prostate cancer produce a protein called prostate-specific antigen, or PSA. Lieber speculates that a sensor could be made of an array of nanowires, each adorned with an antibody selected to latch onto a PSA molecule. An increase in the normal concentration of PSA in the body, therefore, would alter the conductance of the array, which could trigger an alarm warning patients of the biochemical change.

Lieber admits that nanowires may take a long time to become part of technologies available at electronics stores or hospitals. But a hybrid technology that includes microelectronics and nanowires--or other nanotechnology-based components--may arrive sooner, he says. One possibility would be to wedge tiny patches of nanowire-based memories into the minuscule space still available between the transistors of today's chips.

Someday, however, electronics will be based completely on nanowires, says Lieber. "You could throw away your hard disk and throw away everything else and store everything in one little, teeny chip," he says. "Have we done that? No, and there's a lot of work to do to get there. But that's what you can dream about."