A University-led team has found a way to build silicon crystals more efficiently, paving the way for better and cheaper computer chips. See the May 19 issue of Science for more on this research.
Good vibrations lead to better electronics
A University-led team has broken a longstanding barrier to making better, cheaper electronics
By Deane Morrison
May 18, 2006
Like modern-day pied pipers, a team of researchers led by University electrical and computer engineering professor Philip Cohen has "tuned" laser beams to the natural vibrations of atoms and used the lasers to rid silicon crystals of hydrogen--a crucial step in manufacturing computer chips. Because they accomplished this without heating the crystals and introducing defects in their structure, they have broken a major barrier to producing semiconductor devices (like the computer you're using right now) more cheaply and efficiently. Their work is described in the May 19 issue of the journal Science . The researchers have begun the process of patenting the technology. Besides Cohen, Vanderbilt University researchers Leonard Feldman, Norman Tolk, and Zhiheng Liu (who was a postdoctoral scientist at the U when the work was done) along with Zhenyu Zhang from Oak Ridge National Laboratory were part of the team. "We're doing chemistry with a scalpel," says Cohen. "We can break just the [chemical] bonds we want." Anatomy of a silicon crystal In making a computer chip, crystals of silicon are grown layer by layer. Silicon, however, is vulnerable to attack by oxygen--in a process akin to rusting--and so manufacturers routinely protect the silicon atoms by coating them with hydrogen before they are added to growing crystals. But the hydrogen interferes with the bonding of new layers of silicon, so it must be removed before the next round of crystal-building. That means breaking the chemical bonds between silicon and hydrogen atoms, preferably without disturbing any carefully placed wires or charges.
Because they accomplished this without heating the crystals and introducing defects in their structure, they have broken a major barrier to producing semiconductor devices (like the computer you're using right now) more cheaply and efficiently.The heat is on One way to banish hydrogen is to heat the silicon crystals, often to temperatures around 550 degrees C (nearly 1,000 degrees F). But such high temperatures can lead to defects and cut into the yield of chips. Naturally, this lowers the efficiency and raises the cost of manufacture. A better method would be to sever the silicon-hydrogen bond with a laser. This approach takes advantage of the fact that chemical bonds vibrate, bringing the two atoms alternately closer together and farther apart, like the hands of a tiny, ultrafast concertina player. The frequency of vibration in the silicon-hydrogen bond falls in the same range as infrared light. Therefore, if an infrared laser is aimed at the bonds, its photons should resonate with the bond and be absorbed into it. If enough light energy is absorbed, the bond should break and release the hydrogen. In previous attempts at using lasers to break specific chemical bonds, target molecules have quickly absorbed the light energy and turned it into heat. Heat is indiscriminant about which chemical bonds it breaks, and so the weakest bonds break first, leaving the targeted bonds intact. Working at Vanderbilt's W.M. Keck Free-electron Laser (FEL) Center, Cohen and his colleagues took advantage of the FEL's ability to produce light at a wide range of frequencies in the infrared part of the spectum. They found a frequency that cleaved the hydrogen atoms off cleanly--in pairs, as hydrogen gas--even at room temperature. One reason they succeeded when many others have failed may be that silicon is highly transparent to the infrared wavelengths they used. Thus, the light passed through the crystal without interacting with it and so imparted little or no heat energy. The team also found that atoms of a heavier form of hydrogen (deuterium) and atoms of regular hydrogen attached at similar locations on a crystal all exhibited different resonant frequencies. Therefore, the new process can be tuned finely enough to distinguish and select among very similar chemical bonds. "By selectively removing the hydrogen atoms from the ends of nanowires--structures typically five to 10 nanometers [0.2 to 0.4 millionths of an inch] wide--we should be able to control and direct their development, which currently is a random process," says Cohen. Also, the ability to extend a silicon crystal could be used to make airtight seals around minuscule sensors. The possibilities are endless. "This is an example of very fundamental work involving a system that's exceedingly important to technology, and controlling reactions in silicon will no doubt have profound applications," he adds. To Feldman, the Stevenson Professor of Physics at Vanderbilt, the process opens up new vistas in the science of surfaces. "If you stop to think about it, surfaces are where the action is. It is where the rubber meets the road," he says. "In fact, some of the most advanced nanotechnology devices that have been envisioned, like quantum computers, require the level of control that atom-specific processes of this sort make possible."
David Salisbury of Vanderbilt University contributed to this article.
Read a story about this research from the National Science Foundation, which funded the work in part.