Are direct materials consumed in CMP over-specified
such that cost-reductions may be expected, or do tight process windows (i.e.
non-uniformity, dishing, finish, etc.) mandate 5-9s and "ship-to-control" as
usual?
Certainly our goal is to provide the fabs with slurries and pads at lowcost,highquality, and outstandingperformance.As with many manufacturing processes, it is vital to understand the truly critical-to-quality parameters and drive excellence around them.We see this philosophy, "ship to control" as a long term competitive advantage for those suppliers able to do so.This will entail very close partnerships with users of CMP consumable materials to determine those parameters that matter the most.In turn, we drive the same philosophy with our suppliers to drive excellence back through the supply chain.Control-limits tighten with fewer,andmore-meaningful measurements,so it is in both supplier and user interests to really understand and focus on the product features that matter most.This way,we reduce costs by avoiding unnecessary testing while driving quality improvements in those parameters that truly do make a difference.Over-specifying and over-testing serves neither supplier nor user.
Techcet has been following the "ship to control" situation for sometime, as it can be a challenge to suppliers. In the non-CMP areas in the past year, suppliers of materials such as PVD targets to IC fabs have found that the purity demand and "ship to control" has been compromised for cost ... in some cases, considerably lower "9"s. I suspect many believe that "cleaner is better" without necessarily generating compelling data and the current economy may be reassessing this. CMP is one where higher metal impurities may be linked to defects, so perhaps the demands for purity will remain.
CMP pads have typically been underspecified, largely due to lack of understanding of ALL the important parameters that are "critical to quality". A pad spec. sheet may contain only shore hardness and density. So in this case "ignorance" has meant under specification and also contributed to difficulty for customers to adopt alternate pads.
I think conditioners have similarly been underspecified due to limitations in our understanding of all the important performance drivers.
The overspecification may have applied to slurries/chemistries in CMP in the past, in the spirit of "cleaner is better".
My experience in developing and commercializing cleaners (both post-etch and post-CMP) is in agreement with that of Karey and Rajeev in that end users typically believe less is best and over specify the purity of the formulated product. That comes at a cost. Then the end user beats you over the head because the cost of consumables is too high. They do not understand the cost for 6 nines in a formulated product. That may be partly due to over specifying gases, which are infinitely easier to purify compared to that for a formulated product with 6-7 components, some of which are solids. Much work has been done showing that some contaminates are irrelevant; however, that can be very device specific. What metal contaminate kills a DRAM may not affect a CCD and vice versa. This is a vicious treadmill that end users have put the consumable supplier on and I see no easy exit.
We're now (January 20th, 2010) past the
official ending (the 18th) of this virtual roundtable discussion, after 4695 views of 129 replies to 18 questions. I'll
edit together Interesting discussions from most of the topics into a
summary document that will be posted to the Planarization Lounge.
We'll leave the topic posting open in case there are additional comments...but they would not be included in the summary.
Print
