Wafer fab costs may hit $4 billion by 2003
By Lisa A. Karter
Geneva — New fabs built between 2000 to 2003 that will use 0.18 to 0.13-micron geometries on 300-mm wafers are estimated to cost $4 billion, predicts Dr. Alessandro Tonti of SGS-Thomson (Agrate, Italy) in a presentation during Semicon Europa in April.
In his presentation, entitled, “Clean Fabs and Cleanrooms for the 21st Century,” Tonti concludes that by 2010, fabs for 0.07-micron devices may cost $10 billion. The reasons for such “dramatic escalation in expenditures,” says Tonti are increasing mask layers, complicated process sequences and sophisticated tools. In addition, bigger wafer sizes — such as 400 mm — require bigger footprint tools that place a heavy burden on WIP transportation and storage.
Tonti suggests scaling down future wafer fabs using several techniques including roadmapping, or having a well-defined mission statement for a new plant and integrated planning. Several proposals include: aggressive deployment of lot transportation automation, advanced yield management based on a better clean concept of tools, extensive use of in-situ monitoring, and wide-scale utilization of minienvironments.
Another key development in future fabs is the “just clean enough” concept originated by IMEC (Leuven, Belgium) researchers that proposes “not to ask for the very best in purity and cleanliness, but just the level which will do.” According to Tonti, both the user and the supplier are allowed to define the minimum quality level needed without unduly wasting resources.
In addition, Tonti suggests that cleanroom air quality (apart from particles) is “really particle free, but may contain a lot of gases, vapors and impurities suspended in a way that defies filtering.” These includes acid gases and vapors such as HF and NOx; acid mist like H2SO4; vapors such as NH3 and organic amines, dopants, metals and organics. Says Tonti: “There is a growing concern about the potential threat these molecules present to critical surfaces.”
A similar concern exists for organic contaminants — which may come from outside makeup air, from inside process contamination, or from outgassing — in cleanrooms, says Tonti.
Aside from organic contamination, another air quality control issue is the extended use of minienvironments. Tonti cites better air quality control within small enclosures with well-defined air paths and air-wet surfaces as a reason for using minienvironments in the fab. “A lot of work has been done on contamination control in minienvironments, and as a result, available specs are better than current cleanroom air ones. More progress is to be expected in the future with even better materials and more experience in the generation, transportion and deposition of contaminants. In-situ trapping of gaseous contaminants might be possible using adsorbing materials and getters, while particle trapping might be possible by using electrostatic traps and particle confinement by thermophoresis.”
Tonti suggests that plant design developments might include: implementation of zero footprint WIP storage, gradual deployment of full handling automation, same cleanroom size as today or an even smaller one, and other facility changes. “A clever compromise” to these changes, says Tonti, are the “hub-and-spoke” fab layout that could be extended into a three-dimensional layout. The 3-D layout would include a lower cleanroom class, as required by minienvironments, that reduces the need to build fabs with a rigorously planar geometry, Tonti explains.
“The most important point raised, in my opinion, is that of the convenience and necessity of integrated planning on an unprecedented scale for new fabs. The cleanroom will only be the skin of these wonderfully complex structures.”