Concern grows over AMC in lithography bays
06/01/2004
There is increasing awareness about the growing sensitivity of 193nm optics to acidic, basic, and condensable organic species in lithography bay environments. Exposure tool suppliers are developing stringent specifications for acceptable levels of airborne molecular contaminants that cause films on lenses and hazing of optics. Industry measurement and filtration targets are moving from the parts-per-billion to the parts-per-trillion realm in response to growing concerns about airborne molecular contamination (AMC). This article examines AMC trends in photo bays, potential sources for contamination in fabs, measurement trends, and possible solutions.
The meaning of "cleanroom" is being redefined in the photo bay, with growing attention focused on gas-phase contaminants that can condense on critical components in advanced ArF lithography processes. Airborne molecular acids and acidic species (e.g., SO2), molecular bases, and condensable organic species are all suspects in the fight to keep 193nm lenses flare-free.
In response to tightening ambient AMC specifications, semiconductor manufacturers are monitoring cleanroom air more frequently and at lower levels of detection. There is also an increased awareness of possible sources for molecular contamination from high-purity purge gases and compressed dry air (CDA). These gases often are used to keep optical surfaces cool and protected from atmospheric contaminants, but even trace amounts of airborne molecules from these gases pose the risk of creating films on lenses by interacting with the energy of 193nm tools. Greater understanding of contamination levels and mechanisms has even led to an examination of AMC's impact on optics at KrF 248nm wavelengths.
The Yield Enhancement section of the 2003 International Technology Roadmap for Semiconductors also calls for greater attention to wafer-environment contamination control in mini-environments or closed carriers in fabs. Through 2005, the ITRS specifies gas-phase contaminants of <0.75ppbM [equal to <0.75 parts-per-billion-by volume (ppbv) in the gas phase] for NH3, amine, amide, and NMP in the lithography area, primarily for the prevention of chemically amplified resist poisoning and loss of CD control [1].
Through 2009, the ITRS identifies high-purity purge gases for 193nm scanner optics at levels of <1ppbv for species such as O2, H2O ×1000, CO, CO2, and THC. During the same period, the Roadmap calls for critical clean dry air to be specified at <100ppbv for H2O, THC, SOx, NOx, and amines (all ×0.05). No specifications are defined for other species during this period.
Moving ahead of the Roadmap
Faced with mounting evidence of the susceptibility of 193nm optical surfaces to AMC at ppbv and even parts-per-trillion-by-volume (pptv) levels, lithography tool suppliers are apparently unwilling to wait for new ITRS levels to be defined. They are now taking the lead in this area [2, 3].
As more leading-edge semiconductor manufacturers install multiple ArF 193nm scanners in production fabs, new classes of AMC species are coming to light. Whereas "refraction" was always a concern to lithographers, the term "refractory" has come into the photo bay lexicon as well, describing volatile organic chemical species containing silicon, phosphorus, boron, and so on. Refractory compounds, such as siloxanes and organophosphates, have been shown to leave difficult or impossible to remove oxides — termed "refractory" materials by lithographers — on exposed lenses and reticle surfaces [4, 5]. Table 1 lists some of the more common refractory compounds seen in photo bays around the world.
 |
To increase insight into AMC and what roles species play in contamination, analytical service providers and filter makers have teamed up with exposure-tool development groups to move measurement and filtration capabilities from the ppb to the ppt regime [6]. Analytical laboratories are working with chipmakers and industry consortia to develop highly specialized methods and techniques to detect lower levels of potential AMC candidates in an effort to understand the film-formation mechanism [5]. With increasing amounts of data about levels of potential contaminants being collected, the effects of certain species have become clearer.
Also clear is the fact that contamination levels change over time. Spills of solvents such as NMP and developers such as TMAH do occur in facilities; these incidents produce spikes of several hundreds, if not thousands, of parts-per-billion of contamination into cleanroom air. Leaks in process chemical tubing and connectors, even at very low levels, can lead to significant background levels of airborne molecular contaminants. In addition, construction materials can be significant contributors of AMC, especially for condensable organic compounds.
Improvements over time
Figure 1 shows the condensable organic profile of a 300mm volume production fab during the first 16 months of startup and tool installations. A significant drop in condensable organics can be seen during the first six months, primarily due to the effects of decreasing offgassing of solvents and similar compounds from plastics, paints, and other building materials.
 Figure 1. Condensable organic compound levels show a 75% decrease within 16 months after fab startup and tool installations. |
Wafer fab operators certainly are not willing to delay production ramps for months while waiting for condensable organics levels to drop. Device makers also want to extend the life of expensive 193nm scanner optics; therefore, some fabs have installed chemical filters in air-handling systems — at least in the photolithography bay area. Such filters can immediately lower the overall AMC load, and can extend tool chemical-filter life. Room filtration, however, must be analyzed to ensure that it is cost-effective.
Ongoing measurements of cleanroom AMC levels also have uncovered one other interesting trend: cleanrooms, in general, are becoming cleaner. Figure 2 shows data from a representative sample of ambient NH3 levels in more than 150 fabs during a seven-year period. While the data is somewhat scattered, the trend is clear: average NH3 levels have dropped by more than 50%, from 14 to 6 between 1997 and 2003. Much of the improvement can be attributed to modern cleanroom design, control, and filtration. Process-tool emissions also have been significantly lowered by greater attention to exhaust and leaks.
 Figure 2. From 1997 to 2003, measurements show average NH3 levels in a representative number of fab photo bays dropping from just below 15ppbv to around 6ppbv. |
One often-overlooked factor is filtration. Even in fabs without filters in cleanroom air-handling systems, NH3 and related compounds are removed from the air by tool filters. As CDs shrink, more 248nm and 193nm exposure tools with their own chemical filters are installed. This trend is expected to continue as i-line lithography is phased out and chipmakers move to all chemically amplified resist processes.
Other areas to watch
As previously noted, atmospheric airborne contaminants are not the only molecular contaminants scrutinized by exposure-tool suppliers. Compressed gases used for air-bearing stages, pneumatic actuators, and projection and beam-delivery optics are being examined more closely. Even cryogenically produced gases, which are often considered innately pure, are being considered possible sources for AMC. (Six-nines nitrogen is specified to 1ppm. That's 1000ppb!)
Is "contaminated nitrogen" possible? Surveys of cleanroom compressed gas supplies show some interesting results. The good news is that compressed gases are clean, for the most part [7]. On the other hand, some compressed gases are not clean. Table 2 shows results from more than two dozen surveys of CDA and N2 supply lines taken in the photo bay.
 |
CDA is often an overlooked source of AMC. Contamination levels in CDA lines are varied. Condensable organics average <1ppbv (toluene equivalents) and >10ppbv in extreme cases. Clearly, this is a cause for concern when one of the largest users of CDA is the wafer air-bearing stage itself. CDA quality is often a function of CDA-compressor filter maintenance. But once polluted, delivery lines can contaminate for long periods of time. In such cases, point-of-use filtration applied prior to the exposure-tool connection may be a solution.
Cryogenically produced nitrogen also shows some interesting results. In general, contamination is below the lower detection limit of the analytical equipment. This is commonly 0.1µg/m3. If reported as µg/m3 of toluene, the condensable organic concentration should be divided by four to convert to ppbv. So, in all cases, even the minimum levels are <0.1ppbv. If the organic is toluene — which is often not considered a condensable organic in the true sense — there may not be an issue. If, on the other hand, the contaminant is a refractory compound containing silicon, for example, the results would be long-term, continuous exposure of the optics to a potentially lethal agent.
Taking action
There is hope. First, measure, measure, and then measure again. Fabs and tool suppliers need to be able to describe the environment around the wafer and sensitive optics. Figure 3 shows the top five condensable and near-condensable organic compounds by concentration found in the photo bay by traditional sampling and thermal desorption gas-chromatograph mass spectroscopy.
 Figure 3. The top five condensable and near-condensable organic compounds by concentration found in the photo bay. |
With good data, several solutions are possible. Tracking down fugitive emissions can be a relatively easy task. Leaks in wet chemical-delivery systems, leaky or improperly balanced exhausts, and unsuitably ventilated chemical-process tools have shown up as significant points of contamination. More than once, proper ventilation has reduced NMP and related stripper-solvent emissions. Keep in mind that these levels are likely to be well below odor-detection limits and may not be obvious. Real-time monitors are invaluable in tracking these sources.
A second approach is to apply filtration where remediation cannot be applied. Tool filtration to protect photoresist has been the focus of the last decade. Optics protection and the need to control optics-damaging species have taken filtration to the limits of the parts-per-trillion world — a realm both difficult to see and control. Continued measurement of filter performance, ambient cleanroom AMC levels, and correlation to stray light measurements will provide greater knowledge of optics contamination as well as potential solutions.
References
- SIA, The International Technology Roadmap for Semiconductors, 2003 edition, "Yield Enhancements," Inter'l Sematech, Austin, TX, 2003 (http://public.itrs.net).
- K. Lai, C. Wu, C. Progler, "Scattered Light: The Increasing Problem for 193nm Exposure Tools and Beyond," Proc. Optical Microlithography XXIV, Vol. 772, pp. 1424–1435, 2001.
- D. Nam, E. Lee, S. Jung, Y. Kang, G. Yeo, et al., "Effectiveness and Confirmation of Local Area Flare Measurement Method in Various Pattern Layouts," Optical Microlithography XXV, oral presentation, Conf. 4691–4706, 2002.
- T. Bloomstein, J. Sedlacek, S. Palmacci, D. Hardy, V. Liberman, et al., "Refractory Oxide Contamination of Optical Surfaces at 157 nm," Proc. Third Intl. Symposium on 157nm Lithography, oral presentation, Sept. 2002.
- U. Okoroanyanwu, R. Jonckheere, A. Eliat, G. Vereecke, "Experimental Investigation of Fabrication Process-, Transportation-, Storage-, and Handling-induced Contamination of 157-nm Reticles and Vacuum-UV Cleaning," Optical Microlithography XXVII, oral presentation, Conf. 5377-5375, 2004.
- C. Atkinson, J. Hanson, O. Kishkovich, M. Alexander, A. Grayfer, "New Approach to Measurement of Photoactive Deep-UV Optics Contaminants at Sub Parts-per-trillion Levels," Proc. Optical Microlith. XXVI, Vol. 5040, pp. 499–509, 2003.
- A. Mackie, B. Warrick, "Coming Clean about 'On and Off the Roadmap': Sub-100ppt Bulk Gas Supply," Solid State Technology, Vol. 47, No. 3, pp. 35–42, 2004.
Frank Belanger is an R&D chemist at Extraction Systems Inc., 10 Forge Park, Franklin, MA 02038-3137; ph 508/553-3900 ext. 30, fax 508/553-3901, e-mail [email protected].
William Goodwin is VP of manufacturing and engineering at Extraction Systems.
David Ruede is the Filtration Group manager at Extraction Systems.