Machining with Molecules

"This is the heart of the clean room,quot; Bill Mahoney says, as another glass door closes behind us. quot;This is where processing usually starts.quot;

man in clean room suit operating a computer
James Collins

"You sculpt the devices you want:" The Penn State NanoFab houses $23 million worth of tools for laying down and removing materials at the atomic level. The clean-room area is 10,000 times as clean as a hospital operating room.

Mahoney, a staff engineer at Penn State's Nanofabrication Facility, has already brought me through three bright anterooms. In this, the core of the onion, everything looks yellow: my white clean suit, his blue one, the special particle-free paper he's given me for taking notes. The ultraviolet components have been removed from the overhead lights, he explains, because quot;UV is used to imprint the pattern on the wafer.quot;

Beneath our feet is an open grill. Fans in the ceiling push the air straight down, flushing every dust mote, every stray bit of dirt, every skin cell that might escape from my upper cheekbones, the only part of me that stands exposed. This is a Class-Ten facility, 10,000 times as clean as a hospital operating room. The air is changed three times a minute.

Mahoney produces a wafer, a six-inch disk of silicon, its surface polished atomically smooth. Working under an exhaust hood, he sets it on a turntable, then squirts a drop of photo-resist—a red lacquer, like duck sauce—its center. He flips a switch to spin it: 4,000 rpm for 40 seconds, until the lacquer spreads to a thin, even coat. He moves the wafer to a hot plate, bakes it dry, then carefully places it in the photolithography machine. He slides a quot;mask,quot; a clear quartz plate with a white checkerboard in the middle, on top of the wafer, and flicks on the UV light.

quot;The coating under the transparent parts is exposed,quot; Mahoney explains, quot;and light breaks the chemical bonds that hold it together.quot; When the wafer is removed and is placed in a bath of developer, the exposed coating washes away, leaving the checkerboard test pattern. Mahoney dries the wafer with a burst of nitrogen gas and hands it to me for inspection.

quot;What remains of the photo-resist is used as masking through the rest of the process,quot; he says. quot;It protects what's under it while chemical reactions take place in the exposed areas around it. You layer and remove, layer and remove, going from lithography to deposition to etching and back. You sculpt the devices you want.quot;

Computer chips, for instance. Mahoney has just showed me the basics of how Intel and other hardware giants build the dense electronic circuits that add up to the powerful microprocessors that chuckle and chirp inside our laptops and our Palm Pilots. Microfabrication, they call it. But standard microfabrication can begin to seem a little quaint out here at the NanoFab.

In the room next to the photolithography unit, there's a patterning machine that can write lines as fine as 10 nanometers—about 50 atoms—across. It does so, Mahoney explains, with a focused beam of electrons. quot;The e-beamquot; is so sensitive to vibration that it sits on its own concrete foundation—and still has to be shut down whenever the heavy trucks are rumbling at the highway-construction site half a mile away.

In the near corner are a pair of scanning-electron microscopes. quot;We use those to look at features throughout the processing,quot; Mahoney says.

The adjoining rooms contain equally sophisticated machines for laying down and removing materials. Furnaces that heat metals and semiconductors to vapor, which is deposited on a wafer in layers of single molecules. An ion-implantation tool that shoots charged particles below a wafer's surface. An etching tool that works by creating a plasma of reactive particles which chemically attack a surface on an atomic scale. Altogether, the facility houses $23 million worth of state-of-the-art equipment, much of it donated by industry, and a professional staff of 13 engineers.

quot;Twenty years ago, transistors were in the range of 100 microns in length,quot; says Steve Fonash, Kunkle chair professor of engineering sciences and the NanoFab's director. quot;Now, the transistors in a Pentium 4 are at about 100 nanometers.quot; About a thousand times smaller than they were, in other words: 100 nanometers is about one-ten-thousandth the length of the foot of this letter quot;l.quot; In biological terms, that's about the size of a viral particle.

You need some pretty specialized equipment to work with components that small. You need special equipment just to see them. About 20 years ago, researchers at IBM developed the first of these tools: the scanning tunneling microscope.

An STM is like an atomically fine phonograph needle that sweeps along a hairsbreadth (actually, a few atoms' breadth) above the surface of a record. A steady current of electrons quot;feelsquot; the minute gap between the surface and the tip. As the tip scans the surface, this exceedingly sensitive signal is assembled by a computer into a super-enlarged image of the surface topography.

By rendering every atomic feature, the STM opened the door to another world. Once researchers could quot;seequot; nano-sized particles, the next step was to be able to move them around. Remember the front-page articles, with the micro-photographs of the letters quot;IBMquot; painstakingly spelled out in individual atoms? That was ten years ago.

Today, at the NanoFab—and at facilities like it around the world—researchers are working on making it pay: developing the rudiments for components and devices and quot;smartquot; materials that will actually function on the nano-scale.

There's been a lot of talk, fantastical and even fearful, about the immense potential of nanotechnology. The theoretical possibilities that arise out of being able to control matter at so fundamental a level, possibilities for advances not only in microelectronics and information technology, but also in biotechnology and medicine and many other fields, are truly revolutionary. But the working out of most of those possibilities remains in its very earliest stages, with plenty of obstacles still ahead. quot;The doing part,quot; as Fonash says, quot;is not trivial.quot;

quot;Machining at the atomic level,quot; Fonash likes to call it. The metaphor seems particularly apt for the quot;top-downquot; type of nanofabrication that has evolved directly from the pressure on microelectronics to get smaller. But there's another way to build at the nanoscale: Bottom-up. This approach has its roots in chemistry, and taps into the wealth of knowledge that has been gathered over the last half-century about molecular structures and their interactions.

black background with three multicolored rectangles within other rectangles

STM images of nanostructures

One of the most powerful concepts in bottom-up nanofabrication is self-assembly, a phenomenon in which atoms and molecules arrange themselves spontaneously into ordered patterns. quot;You don't have to assemble each individual piece from scratch,quot; Fonash explains. quot;At this scale, that would take forever. What you do instead is set up the right conditions so that molecules will come together on their own in precisely the way you want.quot;

Fonash and his graduate students Joe Cuiffi and Dan Hayes were relying on this process when they recently stumbled on a new type of thin film. To grow the film, they used something called a high-density plasma-enhanced deposition tool, donated to the NanoFab by Lucent Technologies. As Fonash explains, quot;You load the substrate—it can be glass, plastic, metal, foil—into a chamber containing silicon vapor.quot; When a current is applied the vapor becomes a plasma—quot;a very rich soup of highly reactive particlesquot; which, under the right conditions, deposit themselves on the substrate in a characteristic order: in this case as clusters of rod-like columns, like the upright bristles of a hairbrush. Each column, Cuiffi reports, is 10 nanometers in diameter.

The porous surface made up of all those bristle-ends, Fonash says, quot;acts like molecular Velcro, or an English muffin, with lots of nooks and crannies.quot; Those nooks, or pores, are small enough to catch and hold single molecules. And, says Hayes, by adjusting the conditions, quot;You can modify the pores to make them more or less attractive to different molecules,quot; a quality that could one day prove useful for separating out the components of chemical compounds: identifying blood proteins indicative of early-stage cancer or heart disease, say, or trace concentrations of atmospheric pollutants.

Chemist Anat Hatzor took advantage of self-assembly to make what she calls a quot;molecular ruler,quot; a new tool for building surface structures smaller than previously possible. Hatzor, whose office is down the hall from Fonash's, came to Penn State from Israel in 1999 to work as a postdoctoral scholar with associate professor of chemistry Paul Weiss. She was recently appointed chemical domain expert for the entire Nanofabrication National Users Network, the NSF-funded consortium of which the NanoFab is a member. quot;She's the interface between basic chemistry and nanofabrication,quot; Fonash explains.

quot;The object with the molecular ruler,quot; Hatzor says, quot;was to find a way to make very small structures very close together, smaller and closer than you can make with e-beam lithography.quot; Although the e-beam is capable of writing lines as fine as 10 to 20 nanometers, she explains, the closest it can place two parallel lines is 40 to 100 nanometers apart.

Hatzor took this limit as her starting point, using the e-beam to lay down two tiny gold bars on a silicon substrate, with a 100-nanometer channel between them. Then she applied layers of organic molecules, one layer after another, to build up the bars and simultaneously to reduce the size of the gap between them. Here's where self-assembly comes in, she explains. quot;These molecules have a thiol group on the end, and we know that thiol likes to bind to gold. So we just put the bars into a solution of these molecules, and the molecules go and bind themselves to the gold surface.quot;

Because she knows the precise length of her molecules—in this case, two nanometers—she knows that each layer she applies reduces the gap between the gold structures by two nanometers on each side, or a total of four nanometers.

Hatzor adds layers until she gets down to the width she wants, say 10 nanometers. At that point she deposits a layer of gold over the entire structure, creating a cap and filling in the remaining gap. quot;Then we put the whole thing in a solvent,quot; she says, which dissolves both the organic molecules and the gold cap, leaving only the gold that filled the gap and the original gold bars. quot;This part remains because the attachment of gold to silicon is very strong.quot;

What she's left with, in the space between the bars, is a tiny structure 10 nanometers wide and a few microns long: a nanowire. And unlike lithography, which requires that structures be quot;writtenquot; one at a time, quot;this process can create a whole wafer full of structures with one small step.quot;

These tiny structures, mass-produced, might then be taken and used as elements in conventional microelectronic devices much smaller than today's models, Weiss says. But the molecular ruler also represents an advance for molecular electronics, a bustling new field based on the radical idea of using individual molecules to replace the switches, transistors, and other components in traditional microcircuits. Already, researchers have shown that molecules can conduct and store electric current and therefore transmit and hold information. If engineers can find a practical way to connect these individual quot;componentsquot;—using nanowires, or some equivalent—the eventual result could be molecular microprocessors that are tens of thousands of times smaller than anything possible with silicon technology.

The prospect is tantalizing enough that the federal Defense Advanced Research Projects Administration is funding two research groups that are part of a new Center for Molecular Nanofabrication and Devices at Penn State. The Center's members include a dozen academic researchers from Penn State and other universities, as well as industrial and governmental partners including IBM, Lucent Technologies, Motorola, TriQuint, and Sandia National Laboratories. Weiss is the director.

quot;Right now there's a gap,quot; he says. quot;We know how to work from the top down, to scales a little less than 100 nanometers. And we know how to work from the bottom up, from one to a few nanometers. The area in between has been kind of a barren area. One of the goals of the Center is to figure out ways to use simple chemistry—instead of brute force—to hit that middle ground. The molecular ruler is one step in that direction.quot;

Molecular electronics, Weiss acknowledges, quot;is in a very primitive state. We now have molecules that behave as switches, which is essential for logic and memory, but we generally don't know how to hook them up. We don't have the ability to pattern at that scale. Nor do we know what architecture to use to make a useful collection of devices like a computer. Those are all active areas of research.quot;

The work is by nature intensely collaborative. As one piece of the puzzle, James Tour, an organic chemist from Rice University and a Center member, synthesizes molecules that might have useful semiconducting properties—quot;These are phenylene ethynylenes,quot; Weiss says, quot;alternating aromatic rings and triply bonded pairs of carbon atomsquot;—and ships them to Weiss and to David Allara, Penn State professor of chemistry. quot;We hook them up and measure them in operation, figure out how they work,quot; Weiss says.

But quot;hooking them upquot; in any kind of complex, interlinked structure continues to be an elusive goal. Among other things, it means finding the balance in scale between a connecting wire that's small enough and a molecule that's big enough. With the molecular ruler, that balance is closer, but not yet close enough.

quot;Right now, we can make a 10 nanometer wire, and Tour can make a 10 nanometer molecule,quot; Weiss says. quot;but it doesn't do the things we want. If we can get another factor of two or three smaller in our wires, we can move to smaller, more functional molecules. Then we will start to be in business.quot;

Across the street from the NanoFab, in the small-business incubator at Penn State's Innovation Park, are the cluttered offices of the Molecular Electronics Corporation. MEC is a high-tech start-up, and it looks the part: computers large and small dominate three rows of desks jammed shoulder to shoulder; copy machine and coffee pot share a table at the end of the room, and a large white-board full of arcane scrawlings is prominent on one wall. Youthful employees, casually dressed, wander in and out.

MEC is one of perhaps a hundred companies that used the NanoFab last year, and probably the most highly publicized. Stories about the company and its founders have appeared in Scientific American, Wired, the New York Times, and the Economist: pretty heady stuff for a place that only recently expanded its payroll to 11 people. quot;It's because we're the first company that stood up and said we're going to commercialize molecular electronics,quot; says Brad Clevenger, senior scientist.

gloved hands operating machinery, one hand holding a thin tube containing a red substance
James Collins

Chip-making 101: coating a disk of silicon with photo-resist to prepare it for lithography.

MEC was founded in December 1999 by James Tour and Mark Reed, an electrical engineering professor at Yale, along with Penn State's Allara and fellow chemistry professor Tom Mallouk. Two other Penn State faculty, Theresa Mayer and Tom Jackson, both in electrical engineering, serve as consultants.

Though the company started with its operations scattered around the country—quot;We had people in Chicago, in Texas, at Yale,quot; Clevenger says—today, the only aspect of MEC's business that is not conducted at University Park is the chemical synthesis, which is done by Tour at Rice.

quot;This particular high-tech sector is not made for start-ups,quot; Clevenger explains. quot;It's not like starting a dotcom with ten thousand dollars. Ten thousand dollars won't buy you a box of wafers in the semiconductor business. Having access to the NanoFab—and office space at a dollar a square foot—is the only way we can get off the ground.quot;

The field is nothing if not dynamic. quot;Memory, logic, information storage—everything that people were trying to do in the '50s and '60s in silicon is going on in molecular electronics now,quot; Clevenger says. quot;People are just waiting for somebody to find the transistor—the molecule that represents that same sort of epiphany.quot;

MEC's object, he says, is not to replace silicon altogether: that's too big a chunk for any small company to bite off. Instead, the idea is to create some sort of hybrid technology as a first step. quot;Right now we're looking at traditional silicon platforms for different applications and how molecular electronics might fit in.

quot;Some of our biggest problems,quot; he says, quot;come with the counter-intuitiveness of combining these technologies.quot; There's the gross disparity in scale, for one thing. And then there's the fact that organic molecules—as witness the stringent clean-room precautions against contamination by human cells or dirt—have long been seen as totally incompatible with silicon. quot;Developing a constructive organic residue, placing it on a wafer. . . . You have to totally reinvent certain aspects in order to integrate an organic element without its being destroyed by conventional processes,quot; Clevenger says.

As a first step, the company is currently getting ready to hold molecular auditions. quot;We're trying to bring online some simple processes to be able to characterize molecules and better understand what roles they could play,quot; Clevenger explains. Once they have suitable candidates, quot;We hope to identify some application, take a standard silicon platform, and put in our component to complete the process—then demonstrate that the device is smaller, faster, and cheaper than it would have been otherwise.

quot;It's not clear yet that this is possible. Whether our molecular component will actually provide an advantage, and whether that won't be overshadowed by some disadvantage that pops up, there's no way to know.quot;

With a click, Will Hancock starts a video clip on his desktop computer screen. In the jerky black-and-white image, boxcars crawl along a welter of short-line railways pointed in every direction. It's a crowded train yard of nano-locomotives.

quot;What you're seeing is actually axoplasm,quot; says Hancock, an assistant professor of bioengineering at Penn State. quot;Squid neurons. The lines are kind of skewed, because the cell has been squeezed out to put it under the microscope. But you can see the microtubules, and the cargo moving along.quot; Microtubules, he explains, are hollow tubes made of protein polymers, quot;the two-by-fours of a cell. They're stiff, and they help define the structure. They're also the tracks that motor proteins run on.quot;

Hancock is an expert on motor proteins, the enzymes that do the mechanical work within cells. The particular motors he studies are a family called kinesins, which are involved in intracellular transport. Nerve cells, being elongated, have special need for such proteins. quot;The cell body is at one end, the synapse at the other, and they're joined by the axon, which is long and skinny and lined with filaments of microtubules. In order to maintain that synapse, material made in the cell body has to be transported out there. So the cell packages these materials and sends them along. Kinesins are the transporters. They walk along the microtubules to their destination.quot;

woman working in computer in a lab

State-of-the-art Nanofabrication Faciity

The exact workings are still being researched, Hancock says. But enough is known to make kinesins, like other biological motors, of keen interest to nanotechnologists. The scale of these motors—microtubules run about 25 nanometers in diameter—would make them perfect for incorporation in nano-scale devices. Hancock, a member of the Center for Molecular Nanofabrication and Devices, is quot;just getting into the applied research,quot; he says, but already he and Center director Paul Weiss are working on some possibilities.

quot;One idea is to harness these motors to drive a tiny valve or a pump,quot; Hancock says. Both devices would be useful in microfluidics, an infant branch of nano-biotechnology that involves controlling fluids in tiny systems. One widely publicized dream of microfluidics is the so-called lab on a chip, small enough to do analysis on single cells. Another idea, farther out, is an ingestible or injectible nano-device that could travel to trouble spots throughout the body, Fantastic Voyage-style, and diagnose or even treat disease.

In Hancock's envisioning, quot;You might have a plastic or silicon tube with a little piston in it. You could fix motor proteins on either side of the piston, and have microtubules attached along the walls of the tube so that the motors would drive the piston up and down.quot; The motors would be powered via a tiny reservoir of adenosine triphosphate, or ATP, the same fuel that provides energy in cells.

quot;Or take a chip with two reservoirs where you keep different chemicals, and a mixing area in one corner. You could use these motors to open and shut valves, to control which chemicals are released into the mixing area when.quot;

Molecular motors might also be used in a biological version of automated assembly: huge numbers of them could be employed to put together larger structures from simple nanoscale parts.

quot;You would lay down microtubules in a set array on a surface,quot; Hancock says. quot;Then you could attach motors to nanowires or crystals. These motor-coated particles would bind to the microtubules, walk out to the end of the line and stop. In this way you could make a tight assembly of these particles in a two-dimensional plane—a surface that might have some useful properties.

quot;Right now I'm talking to Anat Hatzor at the NanoFab about ways to lay down these arrays,quot; Hancock says. His visions, he realizes, are a long way from fruition. As with most areas of nanofabrication, the promise is far ahead of the actual quot;machining.quot;

quot;People are still working on gaining control over these motors,quot; he says. quot;Getting them to do what we want them to do.

quot;But if we could harness their abilities, it would be something very powerful.quot;

Stephen J. Fonash, Ph.D., is Kunkle chair professor of engineering sciences and director of the Penn State NanoFabrication Facility, 189 Materials Research Institute, University Park, PA 16802; 814-865-4931; sfonash@psu.edu. Joseph Cuiffi and Daniel Hayes are graduate students in engineering science and mechanics. William Mahoney is a member of the NanoFab's engineering staff. Anat Hatzor, Ph.D., is chemical domain expert for the National Nanofabrication Users Network, 194 MRI; 863-8220; stm@psu.edu. William O. Hancock, Ph.D., is assistant professor of bioengineering in the College of Engineering, 218 Hallowell Building; 865-1407; wohbio@engr.psu.edu. Brad Clevenger is senior scientist at Molecular Electronics Corporation, 200 Innovation Boulevard, Suite 226, State College, PA 16803; 814-278-7758; bradc@internetmls.net<.>

Last Updated January 20, 2005