by M. David Levenson, editor-in-chief, Microlithography World
July 17, 2008 – In an interview at SEMICON West, Mark Melliar-Smith, CEO of Molecular Imprints described two paths for his company: one toward insertion into the semiconductor industry in 2009, and another toward producing a billion disks a year with 20nm features for the hard disk drive industry in 2010. He reported that nine tools have already been sold to the hard disk manufacturers (and 20+ to the semiconductor industry), with the Imprio HD-2200 full-wafer tool capable of printing both sides of 65nm disks at a 180/hour clip with <20nm features. Such patterned media will be essential for data storage to maintain its own Moore's law progression (see Fig. 1). For a real factory the lithography tool throughput would need to be 1000 disks/hour, though, he said MII would soon build those tools, based on a cluster tool concept.
While Melliar-Smith does not expect long lifetimes for imprint templates in such an environment, he reported that Molecular Imprints had already demonstrated a low-cost replication technology that employed the S-FIL process with the Imprio-300. Nick Stacey, MII’s director of marketing and business development, noted that an entire generation of R-θ e-beam tools is being developed for patterned HDD media mastering.
Fig. 1: Imprinted disk drive pattern. (Source: Molecular Imprints)
In semiconductor lithography, MII’s roadmap calls for continued improvements in the overlay of its 300 series tools, with 20nm being achieved routinely by the end of this year. That would allow those interested in ≤32nm manufacturing to do process development and integration, observed Melliar-Smith. In 2009, MII expects to develop its first high volume tools, with 20 wafers/hour throughput and 15nm overlay. Then, one year later, those tools would be clustered into a system capable of litho-cell scale throughput. Since the S-FIL tools are so much smaller than steppers and no track needs to be attached, the footprint of even a 10-head system would be comparable to today’s installations, he reported. The cost of ownership at 22nm is projected to be smaller than DUV double-patterning or EUV, though perhaps not as low as today’s 193nm single-pass immersion (see Fig. 2).
Fig. 2: Litho cost comparison at 22nm (Source: Molecular Imprints)
Melliar-Smith noted that neither EUV nor imprint can employ pellicles to protect the mask (or template) from accumulating defects, but that imprint had the “advantage” of being a 1X technology. Thus, the finely crafted master template could be replicated conveniently into low-cost high-fidelity stampers using the proven S-FIL technology. If a stamper was damaged, it could be discarded; in EUV, a contaminated or damaged 4X mask would have to be entirely re-written.
Imprint has prospered in spite of the fact that the majority of litho development attention and funding has gone into other technologies. Now with the need to ramp up production of two classes of tools — a stepper for semiconductors and a full-disk printer for hard drives — progress may be constrained by investment. Molecular Imprints and its R&D partners have demonstrated the viability of their method. Having done that, the time has come to get on with the next (industrial) stage. — M.D.L.