Tabletop EUV Light Source
South Korean research team demonstrates an economical way to generate EUV light using femtosecond laser pulses
9 June 2008—A team of Korean researchers may have come up with an economical way to construct laptop-sized light sources for making the next-generation of computer chips. The team, led by Seung-Woo Kim, of the Korean Advanced Institute of Science and Technology, in Daejon, has demonstrated a method to generate extreme ultraviolet (EUV) light using bow-tie shaped nanostructures and a very fast femtosecond laser. The entire apparatus is small enough to fit on a tabletop, according to their report in the 5 June issue of Nature .
The South Korean method could help to overcome a big problem that threatens to derail the industry’s decades-long march of miniaturization. Transistors today have become so small that a billion of them can be packed onto a computer chip the size of a fingernail. (Today’s state-of-the-art chips have features that are 65 or 45 nanometers in size). These features are etched onto silicon wafers using very-short-wavelength light, but to make faster, more compact circuits, transistors must shrink further. For that to happen, a new, shorter wavelength light source is needed to define the patterns on the chip.
Current plans predict that chip makers will begin to need to switch to EUV for state-of-the-art chips by 2011 or 2012, when the features that make up transistors must be just 22 nm. Finding a suitable light source for EUV lithography machines has proved much more difficult than expected. The South Korean technique is attractive for its small size and optical quality.
In an accompanying commentary in Nature , Mark I. Stockman, a physicist at Georgia State University, in Atlanta, says the South Korean method ”is a deft new way to produce EUV radiation—one that could be considerably more economical than previous approaches.”
Industrial firms working on EUV sources typically blast droplets of tin with high-powered lasers to produce bursts of EUV light. But researchers have been exploring another approach in the lab, which involves starting with pulses of light from a femtosecond laser and stepping up the laser’s frequency. So far, though, the stepping-up process has proved to be cumbersome, costly, and unwieldy. The laser pulses are used to ionize atoms of a gas such as argon, and then the electrons released by ionization are accelerated by an intense electric field. Subsequently, the electrons recombine with the ions, shedding the excess energy as light that is made up of harmonics of the original laser pulses—multiples of the laser’s initial frequency. The highest order harmonic typically falls in the EUV range, and the rest are filtered out.
Kim and colleagues simplified the stepping-up process by generating the electric field needed to produce the high harmonics using a bow-tie shaped nanostructure made up of two cones separated by a small gap. They used a standard titanium-sapphire femtosecond laser that emitted a train of 10 femtosecond laser pulses with a wavelength of 800 nm. The output beam had a modest peak power of 100 kilowatts, and each pulse had energy of 1.3 nanojoules, which gave a modest pulse intensity that needed to be amplified by a factor of at least 100 to generate EUV light. They fed a jet of argon gas over the bow-tie nanostructures, and then they fired the focused laser beam into the gap between the gold nanostructures. The intense electric field produced in the gap by the interaction of the laser light with the metallic points stripped electrons from the argon. When the electrons recombined with their atoms, they produced light of higher order harmonics—up to the 17th harmonic, which was 47 nm, right in the EUV range.
While an EUV light beam was produced, it is still not quite of a narrow enough wavelength or powerful enough for commercial applications.
”The beam is not as powerful and collimated as required for some EUV applications” says Stockman, but he predicts that the ”power will increase” as the system is optimized.
The new method works because of the physics of ”surface plasmons”quasiparticles that form in the gap between the metal cones that make up the bow tie. Surface plasmons form at the junction between a nonconducting material, or dielectric, and a metal. Metals have lots of free electrons, which oscillate when one shines light on them. But when a dielectric such as a gas borders a metal, the movement of the electrons is curtailed because electrons cannot enter the dielectric. When light hits them, the electrons are constrained to move sideways within the surface of the metal, sending an electronic density wave along the surface. In the South Korean experiment, surface plasmons in the bow tie amplify the incoming laser beam and produce EUV light.
The beauty of the method is the economical way in which the EUV is produced. Today’s state-of-the-art lithography machines cost millions of dollars, and chip engineers haven’t figured out a way of reducing the cost of EUV lithography machines, which are forecast to be even more expensive than today’s machines. The South Korean method shows a promising avenue to lower priced and efficient EUV light sources. Not only that, it could be used to produce EUV light for other applications, such as spectroscopy and imaging. ”This new method of short-wavelength light generation will open doors,” says Stockman.
But for industrial use it will have to produce smaller wavelengths of light at higher power. Stockman thinks it can certainly do the former. ”There is a clear possibility to extend the range toward shorter wavelength by optimizing both the nanostructure and the excitation laser pulse,” he says. ”I expect much progress will be made all over the world within the next year.”
Jorge J. Rocca of Colorado State University, an expert on EUV light, who has been developing compact EUV light sources himself, agrees. ”It has the potential to go to the really short wavelengths we are looking for in the future,” he says. However, he’s less convinced that the Korean system’s power level can be increased enough for industrial use. ”It could be challenging to scale to the higher fluxes and shorter wavelengths needed for most applications, as the required drive laser intensity reaches the damage threshold of the materials,” he wrote in an e-mail. ”Maybe clever structures will be able to at least partially mitigate that.”