This site will look much better in a browser that supports web standards, but it is accessible to any browser or Internet device.
by Sally Bell
How would you like to fit the capabilities of your desktop computer into your cell phone? Or maybe wear that cell phone like a wrist radio from the old Dick Tracy cartoons?
The odds are high that these wonders will be a reality in a few years. What will make them possible are microprocessor chips even tinier than those of today, which will allow the gadgetry they power to be smaller, yet faster and less expensive.
But shrinking chips is getting steadily harder. Their individual transistors are already too miniscule to see unaided, while the equally vital copper wires connecting transistors have shrunk to just 100 atomic diameters wide. That's far beyond even microscope scrutiny. The wires now are so vanishingly fine they can only be perceived indirectly via the most advanced scientific armament, yet they are never slender enough.
This is the province of surface scientists, who study chemical structures and reactions at the outermost surfaces of materials and their interfaces with liquids, vapors or other solids. And among scientists working to make the wires in semiconductor chips still thinner, the University of North Texas' Jeffry Kelber, Ph.D., Regents Professor of chemistry and surface scientist, is considered a leader.
|Jeffry Kelber, Regents Professor of chemistry, is working to make microprocessor chips even smaller than those of today. "We all want our gadgets better and smaller and, oh yes, we don't want to pay any more," he says|
Kelber's work centers on how to deposit copper in the infinitesimal "trenches" linking transistors, then sandwich the copper atoms within equally diminutive barrier layers — today's state-of-the-art layer is several hundred atomic diameters deep — so they don't diffuse into the silicon chip itself, where they cause havoc.
"Forming stable interconnects under these conditions requires detailed control of how the atoms form and grow into a film on the substrate surface," Kelber says.
He likens the wire to a highway. Like an interstate highway pocked with large potholes keeps drivers from zipping along smoothly, a wire that isn't properly insulated keeps electrical signals from reaching the transistor and prevents the chip from functioning. Finding the right combination of materials to insulate the copper wires is rather like paving a charged highway.
To make the wires still skinnier and insulate his information highway, Kelber studies the forces governing how copper atoms deposit, adhere and build on surfaces in order to learn how to lay down copper in ever smaller areas. He also studies the chemistry that will keep the metal both chemically and structurally stable under normal equipment operating temperatures.
|The highly uniform arrangement of atoms on the surface of a ruthenium single crystal helps scientists investigate chemical reactions important in microchip fabrication processing.|
Kelber says there is a "practical, realistic limit" to how small chip components can go, limits that have been reached in some areas, but there's still plenty of room for progress in his specialty.
Until just eight years ago, he notes, the wires were made from aluminum. Shifting to copper around 1996 proved a big advance for the industry because copper conducts electricity better than aluminum, and with less resistance. That meant chips could process information faster and could shrink another step — which allowed more chips to be packed into a space. The result is today's small cell phones and powerful laptop computers.
New obstacles, though, seem to lurk behind every advance.
Because copper atoms leak readily into the silicon material of the chip itself, now "not only do you have to put down a very narrow copper line, but you have to surround it by a barrier metal that prevents copper diffusion," Kelber says.
The chip industry today encircles the copper with a layer of tantalum, overlying one of tantalum nitride, because tantalum, a dense metallic element, doesn't combine with copper. Tantalum, though, oxidizes too easily.
That leaves surface scientists like Kelber with a lot of work to do on the next generation of chip barrier material. They want to switch to barriers made of ruthenium, a brittle white metallic chemical element that doesn't oxidize quite as readily and offers less electrical resistance, making copper deposition more reliable and cheaper.
"A lot of what is done in this is trial and error," Kelber says. "We're trying to use our understanding to devise newer and more effective processes."
In fact, this spring Kelber applied for a patent on what he describes as a "relatively simple" way to protect ruthenium from oxidation in ordinary industrial handling. To protect his work, however, he doesn't want to outline his development more openly until the patent is approved, probably next year.
|Kelber studies the forces governing how copper atoms deposit, adhere and build on surfaces. These copper gaskets are used as seals in vacuum systems.|
T. Lloyd Chesnut, Ph.D., UNT's vice president for research and technology transfer, calls Kelber's patent work "a breakthrough that if licensed and recognized" could make the university a national leader in this specialty and advance UNT's goal of becoming a major research institution.
Kelber though, doesn't think in breakthroughs. He doesn't even like the word.
"Progress isn't someone leaping out of the tub and shouting, 'Eureka,' but understanding what you didn't before," he maintains. "Progress occurs through a steady drumbeat of advances that go on almost unnoticed."
"Look what's happened since 1980," he says. "Nobody had a desktop computer then. What's happened in the last 20 years has occurred not from breakthroughs but gradual progression in technology. We all want our gadgets better and smaller and, oh yes, we don't want to pay any more."
Termed a breakthrough or not, the result would be the same. Success in his work — and other aspects of shrinking chips — means "the continuance of what we've been enjoying for the last 40 years: smaller, faster, cheaper computing."
"The benefits would be all things that involve faster and cheaper computers," Kelber says, things like smaller, thinner, lighter and less expensive laptop computers and, yes, palmtop devices as sophisticated as today's laptops. "Our success would greatly benefit the microelectronics industry in Texas. That means lots of jobs for lots of people."
That prospect is why Kelber receives substantial grants from industry. His work is funded by the Semiconductor Research Corp., Austin's Sematech chip consortium and the Welch Foundation.
|An ultra high vacuum system allows scientists to study surface reactions under rigorously controlled conditions.|
Harold Hosack, Ph.D., who directs interconnect and packaging sciences at Semiconductor Research Corp. in North Carolina's Research Triangle Park, says Kelber is one of the relatively small number of surface scientists "willing to do science for the purpose of doing something useful" rather than for pure research only.
For industry, which wants to be first out with the best new technology, Kelber's practical bent is "a very important attribute," according to Hosack.
"The problems the semiconductor industry faces are becoming much more difficult scientifically because the dimensions of materials are very small and the materials more diverse," Hosack says. "That combination makes surfaces and interfaces more important than in the past.
"The things Kelber's doing allow us to understand how to make materials more compatible when joined, allowing us to understand what material it is reasonable to investigate. That helps save both time and money."
J.D. Luttmer, Ph.D., a process engineering manager at Texas Instruments in Dallas, says Kelber's work is "highly regarded" in the chip industry.
"Kelber gives us early learning," Luttmer says. "He helps us understand and fix some of the problems in developing new semiconductor technology."
As small as semiconductor chips are becoming, though, they can't continue their march toward invisibility forever. At some point, quantum mechanical effects must be dealt with. They'll show up, for example, in higher-than-expected electrical resistance and electron leakage even when a device is turned off. That limits battery life, increases power consumption and "causes all sorts of nasty things," Kelber says.
The leakage can't be prevented, he maintains.
"If we are going to continue seeing computing power increasing, this leakage will happen in one way or another," he says.
The answer, then, is making chips that rely on quantum mechanical effects yet mesh with popular gadgetry. But that research, still five to 10 years away, is barely in the thoughts of scientists like Kelber. Today's challenges are tough enough.
"I don't know where all this research will take us," Kelber says. "A lot depends on things other than chip designs. I think we are within sight of the end of the game of making spectacular leaps by making the chip smaller. We'll have to do computing differently or make the devices differently.
"But, of course," he adds, "people were saying that 10 years ago, too."
For more information about Kelber and his research, go to www.chem.unt.edu/faculty/kelber.htm.