Metrology Webcast Looks Into Scatterometry, CD-SEM, the Future

Semiconductor International's second technology webcast addressed the technology that is arguably the greatest enabler for ramping 90 and 130 nm processes in production today: CD metrology. Planned and moderated by SI Senior Editor Alex Braun, the webcast featured three experts in the metrology community: Ulrich Mantz, head of the metrology and inspection department at Infineon Technologies (Dresden, Germany), Alain Diebold, senior fellow of International SEMATECH (ISMT, Austin, Texas) and David Seiler, division chief of the Semiconductor Electronics division at the National Institute of Standards and Technology (NIST, Gaithersburg, Md.). To view the webcast from our archives, go to www.semiconductor.net/webcasts.

Mantz highlighted the new capabilities offered by scatterometry today. "Besides the precision, scatterometry can measure depth, profile angle and shape," he said. "Its high speed allows 100% measurement capability, and in-line integration is possible." The main drawback to scatterometry is, of course, the need for advanced modeling capability, including n and k dispersion modeling and stack modeling, and a model for the optical grating. From there, the user chooses software — using a library or real-time regression analysis. Real-time measurements are compared to the library, or the user performs regression analysis to get the best fit. In terms of speed, library generation can take a day or more, but the data comparison is very fast, Mantz explained. For regression analysis, the setup is fast, but calculation time, which is in real time, can affect measurement time.

Mantz defined scatterometry's five most critical challenges. "Most important is to have an advanced modeling tool and to do the proper assessment of the model," he said. "Also, good process knowledge is required, including knowing what needs to be controlled and what process variation can be expected (e.g., line-edge roughness). You want to make sure that you assess the spectral sensitivity, and that you get the required precision for the parameter of interest."

Scatterometry is typically being used for process control after lithography and etch, at shallow trench isolation (STI), at the gate level, and for litho cell monitoring at critical levels. CD-SEM's advantages include direct imaging, its established role in fabs today, the ability to measure 3-D structures, and small-area analysis capability (Table ). Mantz warned of the dangers of looking for close correlations between scatterometry and CD-SEM. Because of the differences in the techniques, correlation will not be perfect. Going forward with metrology in general, Mantz commented, "One critical point is whether we will have the right precision and sensitivity to meet future requirements for process monitoring, and I think we need to collect more data before we can answer this question."

Regarding in-line integration of scatterometry, Mantz said, "Scatterometry can do a good job of catching yield excursions, but that integration takes some time to implement, which means process tool availability and time for tool-to-tool matching."

ISMT's Diebold foretold the extendibility of scatterometry and CD-SEM. "One of the ways to extend CD-SEM beyond the 45 nm node may be the high-voltage CD-SEM," he said, noting that, in addition to higher voltage, low-loss detectors help sharpen the images (Figure ). "And the perceived damage from higher voltage may not, in fact, play out to be true."

High-voltage CD-SEM (100-200 keV) combined with a low-loss detector greatly enhances the surface detail and reduces edge brightness in this image of copper lines. (Source: International SEMATECH)

Diebold explained that scatterometry may be extended to lower wavelengths, "which may give some assistance with smaller features. But whether it gives better sensitivity in that range needs to be sorted out by the community. The backup technology for these techniques is TEM, which will get more attention over the next several years."

3-D metrology is being developed for scatterometry, CD-SEM, CD-AFM and CD-FIB. Software to convert top-down CD-SEM images to 3-D images is in the R&D stages. Diebold pointed out that high-k gate dielectrics and metal gates greatly challenge metrology because of the precision required and because "right out of the furnace, the high-k material has high interface trap density that causes measurement error in equivalent oxide thickness." Silicon-on-insulator (SOI) film measurement is difficult because "the dielectric function of SOI is not very well known for thin silicon layers (<25 nm)." Diebold added that, generally, interface characterization is a metrology need that is often not met; there is a need to average measurements over large areas to meet precision/tolerance requirements; and there is a need to move toward in-die measurements.

Seiler summarized some of NIST's programs in the areas of DUV lithography metrology (www.eeel.nist.gov/812/omp/lithography.html), CD linewidth metrology (www.eeel.nist.gov/812/41.htm), 2-D and 3-D dopant profiling (www.eeel.nist.gov/812/omp/dopant.html), gate dielectric metrology reliability (www.eeel.nist.gov/812/omp/reliability.html), and porous thin-film metrology for low-k dielectrics (www.eeel.nist.gov/812/omp/dielectric.html). Recent accomplishments include low-k material characterization on porous films, a PC-based program that extracts 2-D carrier profiles from SCM images of doped semiconductors, high-k film investigation with VUV ellipsometry, four new ASTM/JEDEC standards on reliability, and prototype fabrication of traceable CD linewidth reference material with 50 µm long and 40 nm wide features. NIST also recently performed the first C-V measurements on a molecular test structure.