As semiconductor processes head below 65 nm, scientists are scrambling to find a light source that will let fabs manufacture next-generation chips. Martin Richardson, an optics professor and director of the University of Central Florida's laser plasma laboratory, believes extreme-ultraviolet (EUV) light could replace the deep-ultraviolet lithography that's currently used to carve circuit patterns onto silicon wafers.
"We must use a light source with a wavelength that's short enough to allow the minimum feature size on a chip to go down to possibly as low as 12 nm," Richardson says. "The switch to EUV will be the largest technology transition that the computer chip industry will ever make."
UCF's EUV technology directs a laser at liquid microdroplets containing a small amount of tin (see the figure). The laser then heats the target, creating a plasma that radiates light (invisible to the human eye) at 13.5 nm. The light travels into a condenser, which collects and directs it onto a mask that serves as a circuit pattern stencil. Finally, the mask's pattern is reflected onto a series of curved mirrors, which reduces the image's size and focuses it onto a silicon wafer.
Yet a real-world EUV source powerful enough to drive chip steppers—the devices chip manufacturers use to place detailed circuitry images onto wafers—has proven elusive. The argon-fluoride excimer lasers that manufacturers currently use for conventional UV chip lithography simply aren't up to the task for EUV, Richardson says. "There are no strong laser sources at 13.5 nm," he notes.
Richardson and his team turned to Crawley, England-based Powerlase and its diode-pumped solid-state (DPSS) Starlase laser. "It is one of the highest-powered, pulsed solid-state lasers on the market today," says Richardson. The researchers combined the laser with their own tin-doped micro-droplet laser plasma source technology to create an EUV light source that Richardson claims is 30 times more powerful than any previously reported system.
The UCF technology also effectively eliminates the neutral and charged particles emitted by existing EUV plasma sources, Richardson says. If allowed to stream freely away from the source, the particles can harm the expensive optics used in EUV steppers. "This affects the stepper's performance and operational life," Richardson says.
Richardson believes his technology is poised to play a crucial role in EUV-based chip manufacturing. All three major stepper manufacturers—ASML, Canon, and Nikon—have EUV development programs in place and have already developed alpha stepper prototypes, Richardson says. Sematech, the non-profit semiconductor research consortium, forecasts that EUV production will begin at the 45nm level in 2009.
Richardson, however, believes that full-scale EUV manufacturing isn't likely to start in a meaningful way before 2011 or 2012. "Whenever it begins, EUV will be the next generation of computer chip manufacturing," he says. "All that's needed now is a light source, and ours has the required parameters."