quantum leap

As silicon chips are pushed closer to their technological limits, engineers are moving to the molecular level to deliver the next step in computing speed.

By Corinna Wu

Molecules In years past, old folks would gather their grandkids around their feet and spin tales of hardship that inevitably involved having to walk two miles to school uphill  in the snow--both ways. Children of today, living in the information age, might  one day do the same to their grandkids, recalling endless download times with  56K modems and computers that ran at only 500 megahertz.

It's apparent that silicon technology, the basis of modern computing, has changed society immeasurably over the last half century. In 1965, Gordon Moore, cofounder of  Intel, quantified that pace of change. He said that the density of transistors on integrated circuits doubles every 18 months or so, and the speed and power of those chips follows suit. So far, this axiom has held true. Over the years, engineers have repeatedly come up with better ways to pack a greater number of transistors into smaller spaces.

Now, scientists and engineers are hoping to keep this exponential growth from slowing to a halt. To many researchers in academia, government, and industry, the key to fulfilling Moore's prophecy lies with molecular electronics. Instead of carving computer chips out of silicon, these forward-thinking scientists are working toward building  devices from the bottom up out of molecules. Researchers say that these molecules--which are generally elongated organic molecules with electronic properties that allow electrons to flow from one end to the other--can replicate the function of transistors and other elements that carry electric current and store information. Devices based on molecular electronics promise to be not only more powerful but significantly  cheaper.

Chemist James Tour of Rice University explains the potential of molecular electronics this way: "In 40 years of silicon manufacturing, there have been less than [10  million trillion] transistors made. No more than that." Although that sounds like a huge number, "one drop of water has [100 million trillion] water molecules in it. So there's 10 times more molecules in a drop of water than the number of transistors that have ever been made. That's how small molecules are." In a microprocessor, small size translates into speed, since the electrons have less distance to travel. In a computer memory device, small means the ability to pack more information into the same space, boosting storage capacity.

These futuristic dreams have been grounded by some very real and promising results, and the hard  work is starting to pay off, says Mark Reed, head of the electrical engineering department at Yale University. "It has me very excited. Like any technology that's disruptive, it's going to do things we haven't thought of yet." Tour predicts that in 10 years, there will be a working molecular computer. Hybrid systems that integrate with existing technology will appear even sooner--in 3 to 5 years. 

Reinventing the Chip

Not long  ago, the words "molecular" and "electronics" were rarely mentioned together.  Silicon ruled--and still does rule--the world of electronics. But the shiny, silver-colored semiconductor is rapidly reaching its physical and practical  limits. It takes roughly $2.5 billion to build a traditional chip fabrication line, says Tour. The amount of money needed to design and manufacture faster chips "gets beyond even what a consortium can do," he says. Silicon Valley may be awash in money, but even its deep pockets won't be able to leap the technological  hurdles in current chip technology.

The reason behind this barrier is that "silicon is a top-down technology," Tour explains. Transistors are etched into thin silicon wafers using a process known as lithography. Over the years, clever engineers have found better ways to carve out ever-thinner  lines of silicon. But when features get thinner than three atoms, Tour says, the tiny electric currents coursing through the material start to leak out. Silicon acts as a semiconductor because electrons run through a broad swath  of allowed energies known as bands. "When you build a structure too small, the bands start going away," he says. "It's a fundamental scientific barrier, not a technological barrier. Those can always be blown through by good engineers."  It's possible to carve out lines of individual silicon atoms, but they won't  do anything useful.

Also, at that size, quantum effects become more important, making the behavior of electrons less predictable. "If transistors become too small, current can no longer be controlled," says Herb Goronkin, vice president and director of the Physical Research Laboratory at Motorola in Tempe, Ariz. Molecules, on the other hand, don't rely on bands to conduct electrons. Instead, their electrons possess discrete energies--like rungs on a ladder--and can leap from one to the other. Molecules hang on tightly to electrons, keeping information from going astray.

Current manufacturing technology can carve out lines of silicon just five or six atoms thick--uncomfortably close to the limit. Tour recalls a discussion he had in May with a group at Texas Instruments about the three-atom barrier, and how engineers will reach it by 2006 or 2008. "They laughed and said, we'll be there  by 2003 or 2004.'" With this fast-approaching deadline, scientists knew they needed a wholesale reinvention of the technology.Molecules 

History Lesson

The progress in molecular electronics has paralleled that of solid-state electronics, says Ari Aviram of the IBM Watson Research Center in Yorktown Heights, N.Y. The first  modern electronic components were diodes or rectifiers--devices that allow one-way  flow of current. Home radio kits popular in the early 20th century, for example,  were based on diodes made of crystals. Then came triodes--vacuum tubes that  had three electrodes inside. One of the electrodes determined how much current  flowed between the other two. "A very small current in the center one could  control the current between the anode and cathode," Aviram explains. And  vacuum tubes, of course, ran the earliest computers.

In 1948, Walter Brattain and James Bardeen built the first transistor, which essentially turned the vacuum tube into a solid-state device. It was a funny-looking contraption--a plastic triangle covered with a piece of gold foil, suspended by a spring, and lightly touching a hunk of germanium. By 1954, Texas Instruments had begun mass-producing  transistors made out of silicon. Researchers soon began developing ways to connect many such transistors on a single piece of material. These integrated chips, which are smaller than a thumbnail yet contain millions of transistors, power  the computers we know today. Now, the challenge for scientists interested in molecular electronics "is to translate from the solid state to a molecular world," says Aviram.

And researchers have begun to do just that. Like the early electronics pioneers, their first goal was to create a diode out of a single molecule. In 1974, Aviram and Mark Ratner of Northwestern University published a paper proposing the idea of a  molecular rectifier, a molecule that conducts current in one direction. Many scientists in the field trace the origins of molecular electronics back to that event.

It took more than two decades to translate their vision into reality. It wasn't until the 1980s, when nanotechnology became the hot new field, that researchers realized the value in thinking small--and had the tools to achieve their goals. One such tool was the scanning tunneling microscope, or STM, which allowed researchers to visualize and manipulate individual atoms. In 1996, Reed and his coworkers used an STM to measure the electric current flowing across a single molecule.

The amount was impressive. The molecule they tested could transmit a million million electrons  per second, even more than theoretically predicted. Soon, other groups showed  that solid-state devices could indeed be miniaturized into molecules. The first results came from scientists at the University of Exeter in England. They sandwiched  molecules between two metal electrodes and observed current flow in a single direction. But Aviram says the results raised some controversy because the two  metals were different. No one could be sure that the rectifying effect didn't result from the potential difference between the different metals.

Then in 1997, Robert Metzger of the University of Alabama in Tuscaloosa put those doubts  to rest. He and his group successfully synthesized a molecule that channeled  current in one direction even when it was sandwiched between two electrodes  made of the same metal. Since then, advancements in the field have come one on top of another. "This has gone at a speed that's just unprecedented," says  Reed. "It's absolutely astounding. Of course, the venture capitalists want to  know why it's taking so long!"

Now, researchers are starting to build more complicated and practical devices. "This field got a shot in the arm when the UCLA group and ours published results last year," says Reed. "People sat up and said, 'We can really make things. "In July 1999, James Heath and Fraser Stoddart of UCLA announced they had built a switch out of molecules called rotaxanes. Rotaxanes look something like a ring threaded on a dumbbell. The ring-shaped portion can slide back and forth, shuttling between bulky chemical groups on the ends of the dumbbell component. The UCLA researchers sandwiched a thin layer of rotaxanes between two electrodes and showed that the device could act as a switch. Then, by connecting several of these switches together, they fabricated logic gates, which form the fundamental  architecture of a computer chip.

A few months later, Reed, Tour, and their colleagues described another molecular switch consisting  of a thin layer of molecules sandwiched between two contacts. By varying the potential placed across the molecules, the scientists could turn the current flow on or off. These molecules "beat the pants off of silicon today," says  Tour. A silicon switch exhibits a 5 to 1 ratio between the strengths of the on and off states, where a higher ratio means less chance for confusion between when the switch is on or off, thus reducing error. In the molecular switch, the ratio is 1,000 to 1. Then, close on the heels of that achievement, Reed and Tour's group announced the creation of a circuit element that could be used in computer memory.

The dynamic random access memory, or DRAM, in today's computers has to be refreshed 100,000  times a second, says Tour. "Our memory lasts 10 minutes. It's stable over that time." Less refreshing means that it consumes less power. Charles Lieber of Harvard University and his group have also created their own version of a stable memory device.

It consists of a grid of nanotubes, tiny hollow tubes of pure carbon that can conduct electricity.  Each point where the nanotubes cross each other can register an on or off state, determined by how far apart the overlapping tubes are. The device works as nonvolatile random access memory, meaning "whatever information you write stays there," Lieber says.He adds that nanotubes are natural candidates as molecular wires, which will be  necessary for connecting nanoscopic devices together. "This is critical. You can't just have one or two devices. You have to wire these things up." Nanotubes are "the closest thing to a hybrid between a real molecule--that can be synthesized  to a precise structure and can extend for large distances--and the solid-state."

Self Assembely
Self Assembly
 

Big Questions, Big Rewards

With devices like diodes and switches in hand, researchers are turning their attention  to the multimillion-dollar prize: molecular transistors."The rectifier was important in showing that molecules can be electronic components," says Aviram, but the transistor is the component that makes the core of any electronics. "You can build any computer out of it." Tour has synthesized several molecules that fit the profile of a molecular transistor. The tiny size of molecules makes testing a molecular transistor difficult, however. Finding a way to string a molecule between two electrodes was hard enough--"we beat our heads for five years on this problem," says Reed--but a transistor has three contacts instead of the diode's two. Somehow, Tour has to find a way to touch a third electrode to the molecule. "We have whopping big leads," he says. "It's pretty hard to  bring in a third lead."

While researchers in molecular electronics struggle with technical challenges, they are also thinking  about what to do with all these elements once they've built them. "The question is, how do you rig these things up?" says Tour.

The answer is to let the molecules rig themselves up, a process known as self-assembly. Chemists can design molecules that have a natural affinity for each other. Put those molecules in solution, and they will seek each other out. For example, gold is strongly attracted to sulfur groups. By placing sulfur in strategic places in a molecule, scientists can design the molecule to grab onto a gold  electrode or wire.

"One thing that gets overlooked is that [self-assembly] offers a much simpler way of making devices," says Aviram. Making a conventional memory device requires about 500 different steps to carve out patterns in silicon. In molecular devices, lithography could be used just to lay down a pattern of metal interconnects. Then, the chip could simply be dipped into a solution of molecules, which would arrange themselves in the proper places. A manufacturing process like this could decrease the cost of manufacturing a chip to one-tenth of what it is now, Aviram estimates.

Patterns can be written onto surfaces other ways too. Chad Mirkin of Northwestern University and his colleagues have developed a technique they call dip-pen nanolithography. The researchers use the sharp tip of an atomic force microscope to deposit narrow lines of molecules on a surface. The technique works much like the old fountain  pens of yesteryear, with the microscope serving as the pen nib, the solid surface as paper, and the molecules as ink.

Recently,  Mirkin and his group created a "nanoplotter," which writes with eight pens at  the same time, and plan to increase the number to 32. Using software, the researchers can program the nanoplotter to write patterns automatically. "It's going to  be a tremendous discovery tool," says Mirkin, since it allows researchers to generate lots of different patterns in a short time. 

Silicon's Staying Power

Although molecular electronics researchers are gung-ho about the possibilities of the field, none say that silicon is on its way out. "Silicon does a number of things very well," Reed emphasizes. "We don't want to spit into the wind." He and others in the field say that molecular electronics will complement and enhance silicon  technology. With this in mind, they have to face yet another big question: how  to integrate a molecular device into all the silicon ones existing today. In other words, "How do you connect that beaker to the wall?" asks Bill Warren  of the Defense Advanced Research Projects Agency in Arlington, Va., who funds several molecular electronics projects.

Ideally, scientists want to be able to "pull out a chip from a PC and stick another in," says Tour. But as they've learned from their work with molecular transistors, there's a problem with connecting big electrodes to these tiny molecules. Tour says that molecular electronics is fraught with problems, "but so was silicon." The extraordinary success of silicon technology gives these researchers confidence that molecular electronics will follow a similar soaring trajectory.

Another vote of confidence comes from companies that have built their fortunes on silicon. "Silicon will continue to grow and continue to be needed," says Motorola's Goronkin, but "we recognize that conventional silicon technology encounters sizes at which quantum effects become observable. That results in degraded performance." That has prompted Motorola, Hewlett-Packard, and other companies to form partnerships with academics to explore molecule-based technologies.Reed, Tour, and others formed a company in November 1999 called Molecular Electronics Corp. to "go ahead and try to commercialize some of these things," says Tour. Reed says investors are pouring money into it--perhaps the surest sign of acceptance that a new idea gets in today's economy. 

HOW THEY STACK UP

 

Silicon

Molecular Electronics

Speed

1.5 GHz

Much Faster

Size of Key Features

180 nanometers

A few nanometers

Number of Transistors on Chip

28 million

Billions, maybe trillons

Cost of Fabrication Facility

$200 billion by 2015

Perhaps 1/10 that cost

DRAM Electron Trapping Time

A few milliseconds

10 minutes

Unlimited Possibilities

The field of molecular electronics is still young, so no one knows what full-function molecular devices will look like. The first devices will probably use groups of molecules rather than individual ones as elements, says Aviram. Complexity in function will be achieved by connecting these elements in creative ways, he adds.

And there's still a lot to learn about how these tiny devices behave. "Everyone needs to demonstrate that these devices can act in a stable, predictable manner," says Lieber. With some molecules, researchers need to exhibit an extraordinary amount  of control. For example, the particular arrangement of carbon atoms in a nanotube affects its ability to conduct current. "It's not clear how anyone's going to control that," Lieber says. Other unknowns have to do with architecture--the basic manner in which computer elements are organized and integrated. "The architecture for silicon might not be appropriate for molecular electronics," says Goronkin.

But if scientists and engineers can surmount these hurdles, the world may see applications that once existed only in dreams. "If these devices are cheap, small, and even moderately powerful, you have the basis of a body-borne device," says Lieber. "You could have a postage-stamp-sized thing and not have to lug around a laptop."

And Reed is not only excited about what molecular electronics means from a scientific standpoint. He also thinks it could revolutionize thinking among educators. "It's not a traditional area," he says. "You're not going to find it as a subset in any engineering curriculum. Once you do, the field will be mature. What we have to do--our job as educators is--we need to enthuse students to these new, groundbreaking areas. We can say you don't have to follow in the steps of the old fogeys."

Even though these researchers hope to marry molecules and computers in their lifetime, they know that it will be up to future generations to nurture the relationship. 

Corinna Wu is a freelance writer living in suburban Washington, D.C. 

  PER CHIP

Glossary