Reference Metrology Standards: The Looming Nanotech Crisis
Alexander E. Braun, Senior Editor -- Semiconductor International, 3/12/2008
As nanotechnology becomes prevalent,and lithography workarounds like double patterning manuever into the mainstream, it is evident that for crucial measurements to continue providing reliable results, instrument calibration — reference metrology — acquires unprecedented importance.
At the SPIE Advanced Lithography Conference recently held in San Jose, a panel discussion on reference metrology highlighted some of the hurdles faced by device manufacturers in obtaining the metrology needed to calibrate various tools to perform required measurements. One of the moderators, Ronald Dixson, is a senior scientist at the AFM Dimensional Metrology group of the Precision Engineering Division of the National Institute of Standards and Technology (NIST, Gaithersburg, Md.). “A recurring theme in reference metrology and the standards needed is the requirement to address sample-to-sample bias variation and that bias’ dependence on secondary characteristics,” he said, mentioning the dependence of bottom CD in a CD-SEM on the sidewall as the classic example. A perennial argument deals with throughput and accuracy trade-offs in reference metrology. “Reference metrology is perceived as referring to a range of tools that, although they provide greater accuracy, aren’t used for inline metrology because of throughput and speed requirements.”
| Particularly in the area of nanotechnology, the industry is at a point where reference metrology for the qualification and calibration of inline tools is crucial. The problem now lies in referencing reference metrology itself. (Source: NIST) |
While it is useful to have reference tools to qualify or calibrate inline platforms, what references the reference? “Here’s where NIST comes in,” Dixson said. “Obviously, some reference metrology must have a direct linkage to fundamental physical constants. An example is how a high-resolution TEM provides information about the silicon lattice constant. With that intrinsic reference scale, calibration can be accomplished in an accurate, traceable way. With CD-AFM, there’s a need for a standard that makes it traceable, enabling it to provide reference metrology for the top-down SEM.”
Nobody disagrees that nanotechnology would benefit from a good reference metrology program. George Orji, an NIST research engineer currently with Sematech (Austin, Texas) and the panel’s other moderator, commented on this: “Good reference materials or instruments don’t exist for nanotech. And without nanometrology, there’s no nanotechnology. It’s difficult to get a reference metrology source for features being measured at 22 and 18 nm.” This seems especially true when considering some current reference instruments’ limitations, such as the CD-AFM’s; there are questions about how far down the node it can go. The major concern orbits around the fact that techniques that would allow current instruments to be useful at smaller features are not here yet. The TEM is the main reference method currently used to link many tools back to the International System of Units’ (SI) definition of length. This is why a greater understanding of TEM calibration of CD-AFM and the ability to define factors that make for a correct measurement are needed.
A roadblock to the needed R&D and investment in reference metrology is its cost. “Long-term collaboration between metrology providers in pre-competitive research is needed,” Orji said. “Since the reference metrology market is small and the return on investment is equally small, to advance it is necessary to establish close collaboration between manufacturers, governmental researchers and the users of reference metrology.”
Dixson pointed out that 3-D architectures will introduce reference metrology requirements dealing with measurands that are not being considered, defined or treated by reference metrology in general. “There’s a realization that we’re moving away from where a lone CD metric is sufficient; 3-D metrics must be developed and reference instruments’ capability to deliver relevant 3-D metrology will become important.” Although there is a consensus that 2-D measurements are insufficient, 3-D measurands remain an open question.
Double patterning — likely to become a dominant player at the 32 nm node — will underscore the need for strong reference metrology. “Demands on overlay will require that much more attention be given to overlay reference metrology,” Dixson said. “While current AFM technology is adequate for 32 nm node overlay requirements, beyond that there’s a clear need for continued AFM probe advances.”
A global challenge for NIST and instrument suppliers is to keep pace with the industry’s application-specific demands and eliminate the fab-to-fab biases that inevitably develop when solely relying on internal standards. When there is no NIST standard, the prevailing practice is to use whatever is available. The disadvantage is this often results in a stack-up of uncertainties and biases that creep into the process.
While, overall, it is true that there exists a trade-off between throughput and accuracy, there are cases, such as the calibration of CD-AFM tips, where it is best to use a NIST standard with absolute traceable accuracy instead of an in-house version. Another example is strain measurement in films, where demands on throughput are already wrapped up in the NIST standard itself, so that when it is available as a drop-in replacement for the in-house standard, there is no trade of accuracy for throughput.
A snag in all of this is not just that some of these are standards yet to be determined, but the pace at which some are developed. Because of the rate at which semiconductor industry technology progresses, it often happens that by the time a needed standard is developed, the industry is already working on something else — standards development is a moving target. However, considerable work is being done at NIST and by other organizations to determine how to best generate these standards. Meanwhile, new instruments, such as the helium ion microscope, are being studied for various reference applications.
Reference Metrology=Red Bricks?
02/09/2010SPIE: Metrology Must Provide More Accuracy
02/09/2010How CD-SEMs Complement Scatterometry
02/09/2010A Solution in Search of a Problem
07/31/2005CD-SEM: Precision or Accuracy?
02/09/2010
