PORTLAND, Ore. — Patterning 65-nanometer features on chips involves expensive techniques that have prompted leading chip makers like Texas Instruments to begin relying on foundries.

Researchers at the Georgia Institute of Technology and spin-off company Focal Point Microsystems (Atlanta), now claim to have devised a cheaper, easier way to pattern at the 65-nm node. Called 3D multiphoton lithography, the technique still lacks the throughput needed by chip makers today. But within a few years, the researchers hope their process can be scaled down to lower nodes faster and more cheaply than the exponentially increasing cost of scaling down traditional lithography techniques.

"With our technique, we can use polymers to define features on silicon in a manner similar to traditional approaches, but our method provides a much lower-cost way of making such nanoscale patterns," said Georgia Tech professor Joseph Perry.

The key ingredient to 3D multiphoton lithography, according to Perry, who is also a co-founder and vice president at Focal Point Microsystems, is a process called two-photon absorption. By coating wafers with a special polymer called DAPB (di-n-butylaminobiphenyl), a laser can solidify very fine features, since the polymer only becomes insoluble at the precise focal point of the laser.

Scanning electron microscope images of photonic crystal structures shown at (a) high power and (b) low power, with closeups (below).

"The laser cross-links the polymer, which is a process of making bonds between polymer chains that leads to the formation of a highly interconnected network of polymer," said Perry. "This process renders the cross-linked material insoluble."

After etching off the remaining polymer, nanoscale 3D structures remain on the surface, acting as a mask for patterning wafers. Besides fabricating specialized silicon chips, Focal Point Microsystems also plans to use the technique for applications like microstructuring optical surfaces.

While the technique is promising, it won't replace silicon chip fabrication techniques anytime soon, primarily because it is too slow. Focal Point Microsystems is working to speed up the process, which is a throughput issue rather than a material problem, the company claims.

"We already have efficient materials that should allow for fabrication at high speeds," said Perry. "Our biggest technical hurdle now is instrumentation: We need faster instruments to increase the throughput of the fabrication process."

Instead of using expensive electron beams or extreme ultraviolet lithographic equipment, a simple 520-nm laser beam is used to directly write on wafers coated with the DAPB. Since two-photon absorption drops off rapidly with distance from the laser's focal point, only molecules precisely centered on the focal point receive enough light to become insoluble. The trick is that the laser's focal point is much smaller than its wavelength, enabling 65-nm features to be patterned with the relatively cheap laser.

DAPB is about ten times more efficient at two-photon absorption than commercial UV photoactive materials, enabling the researchers to fabricate the 65-nm, 3D patterns.

The technique currently takes about 10 minutes to create a 20- by 20-micron pattern with 30 layers at 65-nm resolution. Only test patterns have been fabricated so far, but the researchers plan to fabricate a compact microspectrometer for telecommunications or sensor applications. They will eventually attempt to fabricate 45-nm features.

The research was funded by the Office of Naval Research, APEX Consortium and the National Science Foundation.

Two-photon absorption is also being investigated by Japanese researchers. Previous published results there have only achieved 100-nm features.