Philip Rack, associate professor in the University of Tennessee's Materials Science and Engineering Department, advances at a brisk clip along a labyrinth of stairways and long corridors leading through UT's Science and Engineering Research Facility.
His destination: Room 101F, a portal to the future of microcircuitry—the technology behind the ever-diminishing size and accompanying growth in power and speed of the computers, cellular phones, personal data assistants, MP3 players, and other familiar accoutrements of the information age. En route Rack passes a number of chambers of discovery where white-coated lab techs peer through safety glasses at the products of their work. Their science, unlike Rack's, is the science of the seen, where the unaided eye often can study the results of experimentation.
Rack soon arrives at a windowless off-white door. A decidedly low-tech strip of transparent tape affixes a message produced on an ink-jet printer. It reads JBX 6000FS/E Electron Beam Lithography System. He swipes his magnetic key through the security slot and enters a dimly lit room, barely larger than a custodial supply closet. Inside, the JBX 6000FS/E electron-beam (e-beam) lithographer is hard at work. Housed in an angular beige-and-white metal case about the size and shape of a commercial air-conditioning unit, the JBX directs a focused beam of electrons at a target—in this case, a silicon wafer a bit smaller than a compact disc—drawing impossibly thin lines on a polymer layer.
How thin? Think well beyond the realm of things that can be observed even with the most powerful optical microscopes. Think beyond the micrometers used to convey the dimensions of the relatively gigantic excreta of a dust mite. Think beyond the diameter of a typical bacterium. Imagine the width of a human hair, then divide by about 60,000, and you're visualizing on the scale of the nanometer—and entering the world in which Rack and his colleagues operate. By the end of a multistage process, the e-beam's lines will become the minute copper wires, insulators, or semiconductors that compose the neural network of a computer's brain.
Were one to assign roles in UT's contribution to this diminutive, but vital, field of science, the dramatis personae from the Department of Materials Science and Engineering would read thus:
Philip Rack, nanoscale carpenter, who uses e-beam technology to create various tools, devices, and materials.
UT Distinguished Professor and Oak Ridge National Laboratory Distinguished Scientist David Joy, e-beam specialist—with the JBX as well as the scanning electron microscope (SEM)—and metrologist, who devises techniques for visualizing, calibrating, and repairing objects at the nanoscale.
UT professor Mike Simpson, innovator, maneuvering deftly at the interface between the physical and biological sciences. Simpson, who, like Joy, holds an ORNL–UT joint appointment, directs the Molecular-Scale Engineering and Nanoscale Technologies Group at UT and ORNL.
The JBX 6000FS/E plays an essential supporting role for all of these efforts. But to date, the tool's most promising contribution involves repairing minute errors that can shut down a production line or result in the creation of scores of defective computer chips, severely restricting the profit flow for manufacturers like Intel.
To create the circuits on a microchip, the lithographic radiation source shoots a beam (electrons, soft X-rays, or extreme ultraviolet light, depending on the system) through a photo "mask"—a flat shirt-pocket–sized rectangular template of transparent quartz intricately patterned by an opaque chrome overlay. The mask guides the beam through a series of lenses that reduces the image size by a factor of 4 and projects it onto the receiving polymer-based resin coating the wafer.
The polymer layer is called a "photo resist," because it responds to the beam exposure either by becoming susceptible to removal by the developing compound (a positive resist—"what shows goes") or by remaining behind after exposure (a negative resist). Once the beams have passed through the mask, the polymer layer takes on the mask "image." Then a dry (plasma) or wet (chemical) process etches away materials that are unwanted and fixes those that will remain in the substrate and become part of a chip's circuitry. A series of masks creates the millions of transistors arrayed in multiple layers—as many as 50—on a typical microchip.
Lithography based on passing a photon beam through a photo mask might still elicit a gee-whiz response from the average computer user, but the e-beam system can also create the masks themselves by writing directly onto a polymer layer and ultimately creating the pattern in the mask's chrome coating.
Consider that creating a typical photo mask used in the mass production of computer chips can cost $100,000 or more. A nanometer-scale defect in one of the mask's patterns can result in a batch of defective chips. In some cases, minute particles can block parts of the mask that should be transparent and allow photon beams or other forms of radiation to pass through (opaque errors). In other cases, material has been scratched off or removed from the mask, allowing light to pass through where it should be blocked (clear errors).
Intel, the world's leading manufacturer of computer chips—and a corporation with obvious financial interest in reducing product defects—has funded Rack and Joy's research on mask repair since 2002.
If Rack shows up for work sporting a nanotechnology tool belt, Joy performs his duties with the equivalent of a nanoscale yardstick—in this case, a scanning electron microscope, which, like the JBX, uses a focused beam of electrons to visualize things at the nanoscale.
"The key problem as we make structures smaller and smaller is answering the question, ‘How big is this?' because if you're going be manufacturing something with it, you have to get that size exactly right every time and prove that it's right every time," says Joy. "So the next step, if you can measure it and you see that it's not right, is ‘Can you fix it?' "
The answer, says Rack, is yes. "A computer program tells the system what it should be seeing as it scans electrons over a mask or other device," he says. Once the system has detected areas that are at variance with the intended design, the researchers employ a technique that removes unwanted particles (in the case of opaque errors) or puts down material (in the case of transparent errors) where, say, there's a break in what should be a continuous line.
In Rack's capable hands, the e-beam lithographer and its complement of ancillary tools can craft nanoscale tools and devices—built one layer at a time—that improve understanding of the interaction among and communication between cells in the human body. Here, where biological and manufactured systems converge, is where you'll find Mike Simpson.
"We're focused on developing a bottom-up understanding of the structure of genetic and biochemical circuits and networks that are central to cellular function," Simpson says. "With Philip's skills at building nanoscale devices, we're creating functional interfaces between biological materials—from individual biomolecules to whole cells—and carbon nanostructures."
In this case, the interface comes in the form of an array of carbon nanofibers protruding upward from a rectangular substrate. Picture a nanoscale bed of nails. Genetic material (DNA) placed on the tips of the fibers can be directly inserted into the information-processing systems of a large number of cell nuclei. "This allows researchers to genetically alter the attributes of the cells and prompt the cells to perform desired functions like producing a pharmaceutically active compound or detecting hazards in the environment," Simpson says.
Through this technique, the introduced DNA is fixed in place and not free to move around within the cell. This prevents the introduced DNA from permanently entering the cell's chromosomes and propagating as the cell divides. Simpson paradoxically terms this a "noninheritable genetic modification," which might someday help allay concerns that genetically modified organisms will run amok.
At Simpson's behest, Rack has also devised fluidic devices equipped with nanoscale channels to study how molecules interact in the concentrated solutions of the human body.
Though Rack, Joy, and Simpson bear no visible signs that might identify them as radicals, they are nonetheless positioned in the leading phalanx of a revolution that is advancing fields as wide-ranging as medicine and agriculture, data processing and textiles, transportation and environmental protection, national security and food safety. With no small irony, nanotechnology is regarded by scientists, manufacturers, futurists, and venture capitalists as the next big thing, and playing a leading role requires a complement of tools—and skills—resident at very few universities.
As Rack puts it, "We're limited only by the tools in our bag." Fortunately for the UT research enterprise, Rack's bag is brimming with useful gear.
Despite its hefty price tag (more than $2 million), relative scarcity (by some estimates, fewer than 20 U.S. research institutes possess one), and precision (capable of drawing lines as narrow as 25 nanometers, the thickness of a bacterium's flagellum), the JBX 6000FS/E Electron Beam Lithography System is based on a centuries-old technology that remains at the heart of most modern printing processes.
Austrian Alois Senefelder invented the process at the end of the 18th century, and his technique used the affinity of oil for oil and the mutual repulsion of oil and water. Lithography—a combination of the Greek word "lithos," meaning stone, and "graph," to write—originally involved use of an oily medium (a grease pencil, for example) to create an image into a limestone slab.
An acid wash then prepared the stone for printing by reacting with both greasy medium and the uncoated stone. The slab was then wet down with water, which was absorbed by the parts of the stone not covered with the greasy medium. An oily ink was then rolled over the surface of the stone; the greasy image picked up the ink, while the surrounding wet areas of the stone repelled it. The inked image on the limestone was then pressed onto a receiving medium, usually paper.
Flash forward a couple of centuries, ratchet the scale way, way down, supplant the limestone slab with a thin polymer layer on a silicon wafer, maintain a constant interior vacuum, and add inordinately complex electronics, a hydraulic cooling system, and a focused beam of electrons to create the image, and, voilà, you've got the JBX 6000FS/E
The JBX 6000FS/E Electron Beam (e-beam) Lithography System is part of a suite of scientific tools that have helped make UT a major player in nanoscale manufacturing, metrology, and manipulation.
"There's a certain set of tools you have to have to construct devices at the nanoscale, and UT is now among the very few U.S. universities that possess that set," says Philip Rack, associate professor in the University of Tennessee's Materials Science and Engineering Department.
Among the tools are a plasma-based chemical-vapor-deposition system (to put materials down), a plasma-based etching tool (to remove materials), and an e-beam evaporator and a "sputtering" system (to apply micro- and nano-thin coatings onto a silicon substrate). "Investments in these tools have positioned us to make significant contributions to the semiconductor industry, one of the nation's major growth industries," says Rack.
Before UT's acquisition of the JBX and the other tools, the researchers had to travel to Cornell University's nanofabrication laboratory to execute their nanoscale experiments. "We were, in a very real sense, working in the dark," says Rack's colleague, UT Distinguished Professor David Joy, an Oak Ridge National Laboratory distinguished scientist.
Even if you've never heard of Gordon Moore, the founder of Intel, or the law that bears his name, you've likely benefited from its proven validity. Simply stated, Moore's law asserts that every 18 months the complexity (that is, the number of transistors and thus speed of processing) of integrated circuits will double, with no appreciable increase in cost to the consumer.
According to Philip Rack, associate professor in the University of Tennessee's Materials Science and Engineering Department, what keeps us on pace with Moore's law is changes in hardware achieved through advances in lithography—the process used to create the minute circuits on a computer chip.
Consider that in 1971 a top-of-the-line integrated circuit contained about 2,300 transistors; Intel is now pushing 1 billion with its latest creation. "We used to be at tens of micrometers for the minimum width of the lines that could be written, but we've consistently reduced that over time," says Rack.
Line width is dictated by the minimum printable size of a lithography system and depends on the wavelength of the light or radiation source and the effects of various deleterious factors like beam diffraction, heat buildup, and warping of the optical equipment.
The wavelength of electrons is measured in angstroms, an even smaller unit of measure than the nanometer, which establishes electron-beam lithography's primacy in terms of drawing very fine lines. Other lithographic radiation sources—soft X-rays with a 1- to 2-nanometer wavelength and extreme ultra-violet light at a 193-nanometer wavelength—are, comparatively, like the broad strokes of a fat-tipped marker compared with the e-beam's fine-point pencil line.
The e-beam's narrow wavelength compensates in precision for what it lacks in speed. While some lithographic systems flood a mask with radiation—like shining a flashlight through a paper cutout and projecting the entire image on the wall—simultaneously exposing a large portion of the receiving wafer, the e-beam lithographer exposes the receiving medium one pixel at a time.
Completion of a complex lithographic process can engage the e-beam tool for a day or more. "It's like the difference between a laser printer that produces an entire page of type in one pass and writing words with a pen—one ink-stroke at a time," Rack says.
While the e-beam system's slow process would prove an impediment to mass production, it's a perfect tool for prototyping, experimentation, and building tiny nanoscale gadgets capable of entering living cells.
— D. B.