Carl Zeiss SMS has introduced the RegC® process for registration error correction on photomasks. In their PMJ paper, they report using the same femtosecond laser technology as in their previously announced CDC32 product, which corrects CDU error. [The PMJ paper was presented at BACUS, since PMJ was cancelled due to the after-effects of the earthquake and tsunami.]

In these instruments, femtosecond laser pulses create local changes in the refractive index of the photomask substrate (in the case of CDU correction) or bulk strain (in the case of registration error correction). Zeiss claims correction of up to 50% of the total placement error contribution of the photomask.

Zeiss also claims that the treated photomasks are stable. Privately they claim that there is no induced flare. One wonders how this is possible after introducing millions of small pockets of color centers and high strain throughout the exposure field.

Further reading:
doi:10.1117/12.896910

doi:10.1117/12.899904

doi:10.1117/12.879885

[Sorry – I neglected to “Publish” this one last fall. – Richard L]

The SPIE Photomask Conference (a.k.a. BACUS) for 2011 recently concluded. Conference attendance was up by 10% over last year. We were pleased to host ten of the papers from the unfortunately cancelled Photomask Japan Conference. All PMJ papers are posted to the web, BTW. The abstracts can be found here.

Highlights:

NuFlare announced specifications for the EBM-8000 mask writer, targeting the 22 nm node. They have coaxed the lanthanum hexaboride source up to a current density of 400 A/cm² and reduced the maximum flash size to 0.35 micron. The field sizes are also reduced, the main field now being 180 microns and the sub-field 10 microns max. They anticipate achieving 2 nm overlay (sparse patterns).

IMS continues to meet their multi-beam development milestones. We were treated to linearity measurements over features from 24 nm to 90 nm (±0.5 nm). So they are able to do quite good multi-e-beam exposures with a fixed stage. The schedule calls for integrating the full sized (512 X 512) blanker array this month, a moving stage and data-path in December, and first mask printing in March 2012.

Lowlights:

In the 1980’s 1X X-ray was the “lithography of the future” and, as one wag put it, “always will be”. EUV looks increasingly like the “lithography of the future”. This conference we were asked to believe that the source vendors are going to leap from < 10W to > 100W by this time next year. I found no believers among the audience. Defects of all manner continue to bedevil mask blank production. It turns out that even a 2 nm bump (or dig) on the base silicon shows up as a defect to EUV radiation. Needless to say, any 2 nm anomaly in any of the 85 or so layers stacked on top also shows up as a defect. To cap it off, the V.P. of Rapid (K-T’s photomask inspection division) decidedly did not commit to fund an inspection tool at this time. (He also subtly hinted that such a tool would be frightfully expensive.)

Oddities:

The squeegee is back. Out of the Insitute for Microelectronics, Stuttgart comes news of an organic 6-bit DAC fabricated by stencil printing. A nice graphic can be found here.

Conference abstracts may be found here for a limited time.