Modest proposals for the ITRS Factory Integration chapter
09/01/2002
If you work in the microelectronics industry, your life generally revolves around trying to guess what the future has in store; and, of course, how you can prepare for it. Occasionally, we guess correctly.
by Mike Fitzpatrick and Ken Goldstein, Ph.D.
 |
A group that has a remarkable record of predicting the future of semiconductor technology is the Semiconductor Industry Association (SIA).
The SIA consists of representatives from chip manufacturers, tool suppliers, universities, laboratories and associated organizations such as Semiconductor Equipment and Materials International (SEMI) and International SEMATECH (ISMT), a global consortium of 12 major semiconductor manufacturers.
 |
The SIA is best known for publishing the International Technology Roadmap for Semiconductors (ITRS), a document that provides a rational "guesstimate" of the technological future of the industry. ITRS at tempts to determine the techniques, production methods, facilities technologies and mea surement methods that will be required in the next 10 to 15 years.
Predicting the future can be a lot of fun, considering that it will be some years before anyone can prove you wrong and the results often arrive on someone else's watch. To see where we are heading, let's look at some recent developments in the 2002 ITRS chapter on Factory Integration and how it may affect your contamination control practices.
ITRS Factory Integration
This critical chapter deals with facilities, gas and chemical delivery systems, ultra-high purity (UHP) water systems, utility consumption, code and Environmental Health & Safety (EH&S) issues, design, construction and tool installation.
The roadmap attempts to predict values for several elements of the production process. For example, the ITRS predicts that by 2010 the typical semiconductor device will have a feature size of 45 nm (0.045 micron) and will be produced on 300 mm silicon wafers. These forecasted values-"metrics"-largely describe the direction that the industry will follow in the next decade. The list of assumptions and proposed metrics includes:
- Product geometries (line widths);
- Critical particle sizes;
- Concentrations of critical particles (number of particles>critical size per cubic meter);
- Wafer diameters;
- Parameters for deionized and ultra-pure water (UPW);
- Parameters for bulk gases;
- Energy and water consumption;
- Construction, move-in and start-up schedules;
- Cost ratios for total capital, facilities capital, installation and materials handling systems;
- Cost ratios for operating costs, total fab operating expenses, utilities expenses and total facilities expense.
The SIA roadmap provides manufacturers with some fairly aggressive goals and alerts them to those areas where technological breakthroughs are required to continue the pace of product development.
Taking the next step
A recent meeting at Arizona State University (ASU) featured a discussion of the future direction of the Factory Integration Chapter. Organized by Dr. Allan Chasey of ASU, the meeting was attended by representatives of major chip manufacturers, suppliers, universities, A&E firms and consultants.
The group developed two significant recommendations: The airborne cleanliness of semiconductor manufacturing cleanrooms should be lowered to somewhere between ISO Class 7 and ISO Class 9; and the purity of the bulk gas systems should be reduced to low-ppm levels, with filtration and purification at those relatively few points where UHP gas is required.
Considering their significant departure from past practice, these recommendations are worth exploring.
The ITRS predicts that product geometries will move to 22 nm (on 450 mm wafers) by 2016. As a result, future production methods will invariably be more contamination sensitive. It has been an article of faith that cleanrooms must become steadily cleaner in lockstep with decreasing geometry size. This may not be the case. That trend towards higher levels of airborne cleanliness is now being reversed by the use of minienvironments.
Last year, in a series of articles in CleanRooms, we explored the subject of minienvironments and discussed the move towards localized contamination control (search www.cleanrooms.com archives). Responding to the increasing use of minienvironments, the current ITRS predicts a transition to cleanroom classifications of ISO Class 6 (FS209E Class 1000) "or worse" beginning in 2004.
The group at ASU took this concept a step further and suggested that the industry move to ISO Class 7 and ISO Class 8 by the end of the decade and to ISO Class 9 conditions sometime thereafter. In other words, cleanrooms will no longer continue to be cleaner and cleaner.
Instead, they will have extraordinary environmental control in the immediate vicinity of the wafer, while allowing the room itself to drift to less-stringent conditions. This requires that process tools and materials-handling systems manage the contamination control requirements of the product, and allows personnel to work in a more relaxed environment.
...and the next
The trend towards increasing purity of bulk gas* systems has been similar to that of air cleanliness. Impurity levels in the late 1970s were in the low parts-per-million (ppm) range with systems constructed of copper and brass components, joined by soldering and brazing.
The copper tubing was "cleaned" but there was no concern with surface finish or internal crevices. Final testing consisted of a bubble-check leak test and particle counting down to 0.5 micron.
Current systems are fabricated of 316L electropolished stainless steel, specially cleaned for semiconductor use, and welded using ultra-pure purge gases. Systems testing includes helium-leak checking (capable of finding leaks that would take many years to form a visible bubble) and particle counting down to 0.1 micron and sometimes 0.01 micron sizes.
The resulting purity levels, in the low part-per-billion (ppb) range, are near universal, and high part-per-trillion (ppt) levels are becoming common. The use of gas filters, with extraordinary efficiencies, has resulted in systems where particles are almost non-existent. As with cleanroom air, this increase in performance came only with significant effort and at a steep price that was reflected in the cost of the gas delivered to the tool and the cost of the distribution system within the fab.
Recently, manufacturers and designers have begun to reconsider this approach. Ultra-pure gas (high ppt to low ppb impurity levels) is required in relatively small flows and at a limited number of locations in the fab. Throughout the remainder of the process, less-stringent purities (high ppb to low ppm) might suffice.
Why not distribute "low purity" gas in "low purity" components? For the majority of applications, high-ppb gas flowing through non-electropolished 316L stainless steel would be more than sufficient. In those select areas that truly require ultra-high purity bulk gas (metallization, plasma etch and a few others), we might filter the gas again and run it through a local area purifier to achieve the required high-ppt to low-ppb contaminant levels.
This would produce gas of acceptable quality throughout the fab at significantly lower price. One word of caution, however: No one has actually tried this approach, and for good reason. Major manufacturers are reluctant to be the first to employ a new methodology in something as critical as high-purity gas distribution. But several manufacturers, gas suppliers and industry consultants are seriously evaluating this approach and see no reason not to use it.
The concepts developed at the ASU meeting are guaranteed to generate discussion, for they propose reversing trends that industry has followed for decades. The ASU group will offer these recommendations for inclusion in the SIA road map. Expect to see them in the near future.
- [Authors' note: When we speak of "bulk" gases, we are talking about those gases that were historically included in this category: nitrogen, argon, oxygen, hydrogen and helium. We are not including those highly reactive gases that have been recently distributed throughout the fab from tube trailers-most commonly silane, ammonia and hydrogen chloride.]
Michael A. Fitzpatrick is program director of microelectronics for Lockwood Greene Engineers. A senior member of the Institute of Environmental Sciences and Technology (IEST), he is chairman for WG012 (Considerations in Cleanroom Design) and WG028 (Minienvironments). Ken Goldstein is principal of Cleanroom Consultants Inc. in Phoenix, AZ, and is a member of the CleanRooms Editorial Advisory Board.