June 12, 2008 – Toppan Printing Co. Ltd. says it has developed the first 32nm generation photomask manufacturing process, one that shortens mask cycle time and “helps control costs,” targeting volume production by June of this year.

The new 32nm mask process involves new materials (blanks) and “significantly revis[ed]” materials and structures, according to the company’s press release. It’s built on new binary type and half-tone type leading-edge maskmaking processes for fine-pattern formulation, improving dimension accuracy in both types, it says. With the new masks, binary-type reticle performance matches the fine-pattern formulation and dimension accuracy of embedded attenuated phase-shift masks, reducing the manufacturing process for critical layer masks and thus shortening photomask cycle times keeping mask costs down.

“Our collaboration with customers, material suppliers and tool manufacturers has produced a new 32nm photomask solution that would have been difficult to develop on our own,” said Toshiro Masuda, managing director and head of Toppan Printing’s semiconductor solutions division. “We will continue to reinforce our alliances with these partners to offer new solutions to our global customers and to streamline and advance semiconductor technology development.”

SST Editor’s Take:

There seems to be much more (or much less) than meets the eye with this announcement, with many important questions initially unanswered.

Michael Hadsell, Toppan’s EVP of worldwide sales/marketing, told WaferNEWS that the process came out of joint development work with IBM at Toppan’s Osaka plant, with several customers having evaluated it, but he wouldn’t identify those customers or disclose the acceptance status. (Still unanswered: Who has a 32nm-capable stepper today, much less running one at volume production levels right now?)

Hadsell also confirmed that the photomasks are being developed for 193nm water-based immersion scanners, which perhaps means EUV is even less likely to be an option for the 32nm node. 193nm (water) immersion litho likely will involve double-patterning, though, which opens up a very large can of worms for maskmakers (e.g., challenges in overlay and defectivity, and non-technical concerns such as extra costs from additional exposures & process steps). He didn’t say whether or how Toppan has overcome any of these — but Franklin Kalk, Toppan CTO, has been saying since at least SPIE 2007 that 32nm double patterning masks could be made using essentially current technology, if only some systematic error sources were adequately modeled, e.g. the position of the beam in the physical lens aperture. (Still unanswered: How did Toppan and/or development partners address and solve such problems with 193nm immersion + double patterning?)

Hadsell also noted that Toppan’s new 32nm masks and process technology “could be applied to non-water immersion lithography processes,” but didn’t speculate when Toppan expected high-index litho technology to catch up. Progress continues with both high-index lenses (e.g. LuAG) and fluids, which might be ready by 22nm — though Nikon’s cancellation of its high-index program sounded a sour note at a SEMATECH’s litho forum last month. (Still unanswered: To what specs does Toppan’s new 32nm masks achieve for non-water high-index immersion litho processes, and how long does it expect to wait before such adoption will occur?)

We’ll keep digging for more technical info about Toppan’s 32nm photomask work, but in the meantime, there still seem to be some important questions yet as yet unanswered. — J.M., D.V., M.D.L.