Let's Move Metrology to the "Value Added" Side!
Alexander E. Braun, Associate Editor -- Semiconductor International, 8/1/1999
So long as
Moore's Law remains in force, metrology will be a principal enabler for the
development and manufacture of future generations of devices.
Users, however, generally do not view it as a 'value-added' factor, but rather as a necessary evil that adds little, if anything, to the bottom line. On their part, metrology suppliers often act as if they had to justify their reason for living and are hesitant to point out some basic realities.
This situation is unsustainable and must be addressed.
We are facing 0.13 and 0.10 µm linewidths and unparalleled high-aspect ratios. At those dimensions, particles as small as 0.09 µm can have disastrous effects. Added to this, with copper's coming, we confront entire new families of defect mechanisms. Preliminary solutions to some of these predicaments are already in the air: processes must be monitored and controlled in-situ, and this will require more sensor-based control -- i.e., more metrology.
Alain C. Diebold, SEMATECH's manager of metrology coordination, once remarked that, 'In an ideal world, metrology capability would be available at the start of process and tool development, and silicon suppliers would have 450 mm wafer-capable metrology tools in time for development of that wafer size.' Diebold recognizes this is not an ideal world and urges the semiconductor community to expand cooperation.
This is not happening.
Device manufacturers are reticent to reveal their big picture to those producing the metrology tools they need, confining themselves to handing out specs of what they need (or think they need) to their metrology suppliers. 'Partnering' is a tired word that nevertheless applies in this instance. If IC manufacturers expect to continue leapfrogging successfully from one generation to the next, they are going to have to partner -- proprietary information notwithstanding -- with their metrology suppliers. This is not new; the fabless semiconductor industry has been doing it, profitably, for years with their third-party fabs.
On the other side of the equation, metrology manufacturers must push for an active part in device evolution, not just meekly accept a list of specs. They must ask hard questions. Often, IC producers demand capabilities they do not need because they look good on paper. And some of these demands approach theoretical physical limits.
Both sides must talk to establish what must be measured and monitored, what should be measured and monitored, and what would be nice to measure and monitor. For example, design engineers are blithely speccing materials with ppt purities, mostly because they think it might help a balky process. Few in metrology are prepared to stand their ground and point out that reliable, precise measurements at those levels are well-nigh impossible and that, instead, better process monitoring and control would bring faster, better and more cost-effective results.
It is time to recognize that metrology belongs on the 'value-added' side of
the ledger, and that detailed information and ideas must be exchanged freely
between device manufacturers and metrology suppliers if we are to continue being
one of history's most successful industries. 
