32 E-Beams Better Than One

Electronic News, April 17, 2000 by Jeff Chappell

As the semiconductor industry begins to transition to 193-nanometer lithography, one Silicon Valley company feels it has a beadmake that beam, as in electron beamon the future of lithography.

Ion Diagnostics Inc. is working on developing e-beam technology for maskless lithography. While the use of electron beams, columns, and direct write or maskless lithography is not new concept, Ion Diagnostics' prototype machine uses multiple beams per column with multiple columns 32 beams per column, for a total of 6,000 beams focused directly on the wafer.

"It's just brute force," said N. William Parker, Ion Diagnostics' president. The company expects to have a working prototype later this year.

Parker doesn't expect the semiconductor industry to beat a path to the Santa Clara, Calif.-based company's door just yet, however. "No one is going to go away from optical lithography unless they have to," he said.

But he does envision a market niche for Ion Diagnostics maskless lithography tool. Semiconductor devices are shrinking in size and driving up photomask and other production costs; a new 300mm fab will cost anywhere from $2 billion to $2.5 billion to build. At the same time, demand for consumer devices is driving a corresponding need for custom devices that are produced in relatively small amounts. Removing the cost of complex masks would make the production of custom devices much more economically feasible.

"There is a huge cost of ownership advantage for ASIC and system-on-a-chip makers," Parker said. He also suggested that by eliminating the mask steps, IC makers could bring devices to market much faster as well. He put forth the idea that maskless lithography may even be "final-generation lithography."

Other executives involved in the lithography industry agreed that maskless technology has some merits, at least in theory.

"The idea, of course, is really appealing, particularly for ASIC suppliers," agreed Moshe Preil, director of strategic marketing for lithography module solutions for KLA-Tencor Corp., San Jose. Among the giant metrology company's product offerings is reticle and mask inspection. Although Ion Diagnostics multibeam, multicolumn approach may give the company an edge over competing maskless technologies, the industry is still looking at a very long development cycle, Preil said.

Parker suggested that next-generation lithography technology probably won't begin to appear in any significant volume until IC geometries shrink to 0.05 micron. Although Ion Diagnostics wouldn't necessarily have to wait for that technology node, according to Parker, the company doesn't plan to swim against the conservative semiconductor tide. Rather, Ion Diagnostics plans to market its product as a wafer inspection tool for the time being.

Using e-beams for inspection is nothing new, and Parker figures using his company's tool as an inspection tool will get the new technology in customers' fabs and give them a chance to become comfortable and familiar with the technology.

Side-By-Side Technologies

Bill Volk, director of marketing for KLA-Tencor's reticle inspection division, suggested that while maskless lithography could save manufacturers money, it doesn't eliminate all the costs associated with masks. Much of the die-to-database verification insuring that the circuit designer's original design coincides with what is being put on the wafer takes place during the mask inspection phase. Take away the masks, Preil and Volk said, and that verification has to be done at the wafer level.

By converting the pattern to electronic data used by a multibeam electron lithography tool and verifying it with software against the original pattern data, "we argue we can make it even more reliable," Parker said. He added that electronics consistently improve, which will only improve the reliability of maskless lithography.

Ion Diagnostics maskless lithography may have a future in IC manufacturing, but it probably won't completely replace optical or photolithography, or even other e-beam technologies. Preil suggested that in the next few years as 193nm lithography becomes more prevalent, different lithography technologies may be more suitable for different IC manufacturers, depending on their product. Whatever technology prevails in next-generation lithography, it may or may not be economically viable for all manufacturers. "Based on what we know today, it looks like we are heading into multiple generations," he said.