The quest for smaller, more powerful computer chips advanced recently with the discovery of a new technique for processing ultrahigh-density semiconductors.

The new method, utilizing powerful laser blasts in lieu of a conventional heating process, is being developed at the Naval Research Lab (NRL) in Washington D.C. in a collaborative effort with physicists at Sam Houston State University. An article detailing the scientific breakthrough was published in the Feb. 24 edition of Physical Review Letters, a scientific journal chronicling new discoveries in physics.

The finding could ultimately lead to a new generation of smaller, more powerful and energy-efficient microprocessors--the central component of personal computers, according to David Donnelly (left) and Billy Covington, two SHSU physicists working with the NRL team.

Currently, microprocessors, such as Intel's popular Pentium chip, are produced from a single crystal of silicon which is manipulated by adding materials, called dopants, that transform the silicon's electrical properties creating miniature transistors. These dopants are introduced to the silicon by a high-speed beam that penetrates the surface.

During this process, however, the silicon crystal is damaged.

In order to repair this structural damage and electrically activate the dopants, the implanted silicon crystal must undergo a process called "annealing."

"Basically, the annealing process involves taking atoms from where they happen to be, and putting them in the places you want them to be," Donnelly said. "To do that, you have to introduce some form of energy into the (silicon) crystal."

The traditional annealing process involves heating the dopant-implanted, or "doped," silicon in high-temperature ovens. While this process works well, he said, it creates problems.

"If the transistors on the chip are very small, the heating process tends to diffuse them, smearing them out."

This effect limits the size and density of transistors implanted on the silicon crystal, and ultimately, the size of the computer chip as well.

The thermal annealing process also limits the number of chips that can be produced from a single silicon wafer. Right now, manufacturers use eight-inch wafers from which several chips are cut after annealing. While larger wafers would render more devices, Donnelly said, they are too big to allow the uniform heating needed to complete the annealing.

These problems and more are addressed by the new annealing technique being developed by SHSU and NRL researchers.

Rather than ovens, the new non-thermal process employs laser beams to repair and activate the doped silicon crystal.

"When the high-energy pulse of radiation hits the surface," Donnelly explained, "we believe it launches a shock wave that bounces around inside the silicon crystal and kind of moves the atoms around and gets them where they need to be for everything to work properly."

"It's just like hitting it with a hammer," Covington added. The atoms are literally rattled into proper alignment, correcting the defects and activating the chip's electrical properties.

Because the non-thermal annealing is accomplished in a very short time without diffusing the transistor profile, Donnelly said, the new technique allows for the fabrication of much smaller, more densely packed transistors, and ultimately, smaller computer chips requiring less power to operate.

Because electrical currents generate heat which can ultimately damage semiconductor devices or limit their capacity, Covington said, those requiring less power to operate run cooler, longer and more efficiently. If less heat is generated, more transistors can be packed into a smaller chip.

Non-thermal annealing could also alleviate the problems associated with heating larger silicon wafers, Donnelly said, setting the stage for fabrication of more chips in considerably less time. Where the thermal process requires the silicon to bake about an hour in a high-temperature oven, the non-thermal method takes only a five-nanosecond blast from a pulse laser--that's five one-billionths of a second.

The idea to apply laser technology to semiconductor fabrication arose quite serendipitously one Christmas past, when Covington and NRL researcher Charles Manka, former chair of the SHSU Physics Department, were talking shop during a holiday visit in Huntsville, Texas, where SHSU is located.

"We were sitting around talking about two different projects and we saw that there was a point of convergence," Covington recalled.

Manka had been using the NRL's powerful pulse lasers to study the effects of shock waves created by nuclear explosions. Laser blasts, the physicist found, can simulate these cataclysmic reactions on a small scale.

Covington, a solid state physicist, had been studying the properties of silicon and the problems associated with the annealing process.

"Manka had been banging his laser into some materials and watching what happens and I was spending a lot of time doing heat treatments on my silicon samples to fix problems," Covington recalled. "Between the two of us, the idea of using his laser to bang my samples emerged."

"It was one of those things that I'm not sure anybody expected to work," Manka said, "but it was an interesting question and one that experiment seemed better equipped to answer than theory."

When Manka returned to Washington, he zapped a few of Covington's silicon samples with the NRL's pulse laser and sent them back to Huntsville.

"I looked at them and noticed an improvement in their properties, so the rest is history," Covington said. "We started pursuing it from that point on."

In spite of their success with the first laser blasts, it was actually a few years later before a scientific team was assembled to investigate the discovery.

"It just kind of sat there for a while," recalled Manka. "We weren't sure what to do with it."

The project got "an infusion of new blood," he said, when SHSU physicist David Donnelly "came on board" and helped Covington convince the NRL to pursue the research.

"Their (the NRL's) attitude, I guess, was 'if you think it's so interesting, why don't you tell us what to do," said Manka.

And that's exactly what happened. Bolstered by funding from the Defense Advance Research Projects Agency, the NRL formed a multidisciplinary research team to tackle the project.

"We got very enthusiastic about it," Manka recalled.

The article appearing in Physical Review Letters details the results of the team's year-long effort which has culminated in a patent application for the non-thermal annealing process. The patent names Covington, Donnelly, Manka and Jacob Grun, the NRL team leader, as developers of the new technology.

So far, the researchers have effectively applied the non-thermal annealing technique on silicon wafers that have been doped, or implanted, by a process called "neutron transmutation." In this process, the silicon crystal is bombarded with neutrons and a nuclear reaction ensues, causing the silicon to change to phosphorus.

The next phase of research will employ the new non-thermal annealing technique on ion-implanted, industrial scale wafers. Ion implantation, the current process used to fabricate computer chips, involves bombardment of the silicon crystal with positively charged atoms, or ions.

In the second stage of the project, the NRL will be working with ion-doped samples provided by computer chip manufacturers, Donnelly said. An effort is currently under way to secure funding from the National Science Foundation for this next crucial phase of development. As soon as the funding is in place, he said, the research will resume.

If the non-thermal annealing process can be made to work on the industry standard, ion-implanted wafers, Donnelly said, a new era in computing could arise. Such low power semiconductor devices are a key technology in meeting the military's needs for higher computational capability in compact, portable, battery-powered packages with minimal cooling requirements.

Civilian applications for the new generation of ultrahigh-density semiconductors, he said, are virtually limitless, but a great deal of work remains to be done before the technique is ready for industrial application.


Media Contact: Phillip Rollfing

Feb. 21, 1997