SEM Measurements Cause 193 nm Resist Slimming
Aaron Hand, Managing Editor -- Semiconductor International, 12/1/2000
 |
| Aaron Hand,
Managing Editor |
Line slimming caused by scanning electron microscopy (SEM) was a popular topic of concern at Interface 2000, a lithography conference held last month in San Diego. Lithographers have been discovering that — although their 193 nm results may be relatively good — the SEMs of those results are showing a different picture.
About a year ago, International SEMATECH (Austin, Texas) purchased a full-field 0.6 NA 193 nm scanner to explore manufacturing issues at the 130 nm node and extendibility to the 100 nm node. Aware of complaints about ArF resists, the researchers chose Sumitomo PAR-710 as the most appropriate resist. They used 390 nm of the resist on 60 nm of Brewer DUV30 bottom anti-reflective coating (BARC), without optimizing the BARC process for 193 nm lithography.
Although they saw good results in the factory, the results were terrible after they shipped the equipment to International SEMATECH's facility, said James V. Beach, project manager of SEMATECH's Advanced Technology Development Facility (ATDF) 193 nm lithography program. The resist profiles at the source factory showed relatively smooth sidewalls and rounded tops, but the profiles after installation at SEMATECH showed increased sidewall roughness and some T-topping.
After some investigation, the researchers discovered that — rather than differences in lithography processes at the two facilities — the difference was in the SEM process. Fixes for this situation are expected to come from both resist suppliers (with improved resist properties) and metrology tool developers (with more limited e-beam exposure), Beach said, but he stressed that CD measurements should be made conservatively and consistently to ensure valid results. He also discussed the possibility of alternative imaging solutions such as scatterometry or atomic force microscopy.
Researchers at Arch Chemicals Inc. (North Kingstown, R.I.), Honeywell Inc. (San Diego) and KLA-Tencor Corp. (San Jose) studied the effects of SEM measurements on 193 nm resists with a view to improving the measurement methods. What they found was that the more they measure the resist lines with CD-SEM tools, the smaller the lines get. Surprisingly, however, they did not see similar slimming effects from e-beam curing, said Tom Sarubbi of Arch Chemical's Semiconductor Photopolymers Division. In fact, the e-beam curing helped to stabilize the CD.
The e-beam radiation of the SEM process causes ArF resists to cross-link and lose mass. The resist shrinkage that results can be controlled by reducing e-beam dose and giving the resist time to cool during the process.
A group at the Commissariat à l'Energie Atomique/ Laboratory for Electronics, Technology and Instrumentation (CEA-LETI, Grenoble, France) and STMicroelectronics (Grenoble and Crolles, France; and Agrate Brianza, Italy) has studied the problem as well. Experimenting with an ASML PAS 5500/900 lithography system and a resist thickness of 390 nm, the researchers studied several different ArF resists. Their studies showed that the COMA (cyclo olefin alternated maleic acid) resists were about 40% less sensitive to e-beam radiation than the acrylate resists.
The researchers concluded, however, that they could not totally get rid of the resist shrinkage, said Laurent Pain, who is in charge of resist process development for 193 nm lithography in CEA-LETI's microelectronics department. There is always a trade-off between slimming and SEM image quality.
The most effective solution lies with the resist developers, according to Bo Su, a senior technical staff member in Applied Materials Inc.'s CD Metrology in the Process Diagnostics and Control Division (Santa Clara, Calif.). As he notes, shrinkage was also observed in 248 nm resists before suppliers modified the formulations.
But Su and Andrew Romano, product manager of advanced products at Clariant Corp.'s Electronic Materials Business Unit (Somerville, N.J.), also found other ways to reduce the shrinking phenomenon. They recommend using lower-energy electrons and lower e-beam doses. One way to achieve lower doses is to focus the SEM off-site. •
For additional information on lithography news, go to www.semiconductor.net/lithography.
Talkback
Related Content
Related Content
SPONSORED LINKS
