PFCs in the Semiconductor Industry: A Primer
Dr. Joe Van Gompel BOC Edwards, Austin, Texas -- Semiconductor International, 7/1/2000
 |
At a Glance | | |
|
| ||
| |||
Global warming — the science
The vast majority of the earth's energy comes from the sun. The energy flux of sunlight at the earth's surface is typically 250-400 Watts/m2. As a blackbody source, sunlight emits the majority of this light in the visible and UV regions of the atmosphere, 250-800 nm, where energies are high enough to cause interactions in the electron shells of molecules. However, significant levels of energy are still present in the infrared region (2000-20,000 nm, or 2-20 µm). It is in this region where the vibrational modes of molecules are active, so absorption of this frequency of light results in thermal excitation, or heat. Known as radiative forcing, rocks, water, clouds, plants, sunbathers, etc. absorb some of the sun's light, and the rest is reflected back into space. Much of the sun's energy has been locked up for millions of years in oil and coal and is now being released for energy through combustion.2
The atmosphere also absorbs a portion of the IR energy from the sun (Fig. 1). The infrared spectrum of the atmosphere is marked by strong water absorptions at 1400-1800 reciprocal centimeters (cm-1, or 5.5-7 µm) and also at 2.7-3.5 µm (not shown). CO2 also shows a significant absorption centered at 2300 cm-1 (4.3 µm). However, the atmosphere has a significant gap at 800-1200 cm-1, which ordinarily would be absorbed by the earth's surface or reflected back into space. Coincidentally, it is in this region where the absorbances of PFCs are strong, as indicated.
About two-thirds of the infrared absorbances in the atmosphere are due to water vapor, which keeps the Earth from getting cold — as happens on Mars, where water is scarce. Of the remainder, about 81% of the global warming emissions in the U.S. are due to CO2. Atmospheric levels of CO2 are on the order of 500 ppm, up from about 280 ppm in pre-industrial times. After CO2, nearly 10% of global warming is from methane, 6.5% from N2O, and about 2% from PFCs.3 The contribution of PFCs due to the semiconductor industry is about 5% of total PFC emissions, or about 0.1% of the U.S. total of global warming emissions.
| Table 1. U.S. Global Greenhouse Gas Emissions, 1998 (MMTCE)3 | ||||||
| Gas/source | 1990 | 1992 | 1994 | 1996 | 1998 | % |
| CO2 | 1340.3 | 1350.4 | 1404.8 | 1466.2 | 1494.0 | 81.4 |
| CH4 | 177.8 | 179.4 | 181.5 | 183.0 | 180.9 | 9.9 |
| N2O | 108.1 | 113.1 | 121.4 | 121.4 | 119.2 | 6.5 |
| HFCs, PFCs, SF6 | 22.3 | 23.5 | 25.1 | 33.5 | 40.3 | 2.2 |
| TOTALS | 1648.5 | 1666.8 | 1732.8 | 1804.1 | 1834.4 | |
The emissions of global warming gases are commonly reported in million metric tons of carbon equivalents (MMTCE), which is a way of comparing emissions based on CO2 (Tables 1 and 2). The MMTCE formula incorporates the Global Warming Potential GWP100 as shown in Table 3 along with the conversions shown in the following equation:
MMTCE = (PFC (lb) x GWP100/109) x (12/44) x 0.4536
where:
•PFC = specific Perfluorocompound
•GWP100 = value from Table 2
•109 = conversion from grams to metric tons
•12 = atomic weight of carbon
•44 = molecular weight of CO2
• 0.4536 = kg/lb.
| Table 2. Emissions of HFCs, PFCs, and SF6 (MMTCE)3 | |||||
| Gas/source | 1990 | 1992 | 1994 | 1996 | 1998 |
| Subs. of ODS’s | 0.3 | 0.4 | 2.7 | 9.9 | 14.5 |
| HCFC-22 production | 9.5 | 9.5 | 8.6 | 8.5 | 10.9 |
| Electr. transmission | 5.6 | 6.2 | 6.7 | 7.0 | 7.0 |
| Magnesium production | 1.7 | 2.2 | 2.7 | 3.0 | 3.0 |
| Aluminum production | 5.4 | 4.7 | 3.2 | 3.2 | 2.8 |
| Semiconductor mfg. | 0.8 | 0.8 | 1.1 | 1.9 | 2.1 |
Table 3 shows several PFC gases currently in use in the semiconductor industry, along with their atmospheric half-lives and global warming potentials. The global warming potential of CF4, for example, is 6,500, which means 1 liter of CF4 will absorb 6500 times more heat from the sun over a 100-year period than would 1 liter of CO2. When taken over its significantly longer lifetime, CF4 absorbs about 850,000 times more IR energy than CO2. This is the basis for using MMTCE for comparisons between emissions of different global warming gases. Since CF4 is so long-lived, gases released today will be in the atmosphere essentially forever.
| TABLE 3: Global Warming Potentialsand Atmospheric Lifetimes3 | ||
| Gas | Atmospheric lifetime (years) | GWP* |
| Carbon dioxide (CO2) | 50-200 | 1 |
| Methane (CH4) | 12±3 | 21 |
| Nitrous oxide (N2O) | 120 | 310 |
| Halocarbon 23 (CHF3) | 264 | 11,700 |
| Halocarbon 14 (CF4) | 50,000 | 6,500 |
| Halocarbon 116 (C2F6) | 10,000 | 9,200 |
| Halocarbon 218 (C3F8) | 2,600 | 7,000 |
| Octafluorocyclobutane (c-C4F8)4 | 3,200 | 8,700 |
| Octafluorocyclopentene (c-C5F8)4 | 1 | 90 |
| Nitrogen trifluoride (NF3) | 740 | 11,700 |
| Sulfur hexafluoride (SF6) | 3,200 | 23,900 |
| *100-year integrated values. | ||
Global warming and world politics
In 1988, the Intergovernmental Panel on Climate Change (IPCC) submitted information on the anthropogenic influences on world climate. The IPCC, a worldwide organization of 2500 scientists, is accepted by the international community as the authoritative body in this field. Updates from the IPCC continue.
Global warming has stirred much debate in the U.S. government. While reduction of all global warming emissions (most notably CO2) is seen by some as imperative for the future, others are unsure of the science and associated predictions. The latter viewpoint leads to questioning of the potentially significant financial and economic repercussions of forced reduction of emissions of CO2 and related compounds. As a consequence, any legislation pertaining to emissions reduction is subject to intense consideration and scrutiny.
In 1992, the Framework Convention on Climate Change, better known as the Rio Summit, set forth initial policies to reduce emissions of global warming gases. The policies were all voluntary, kind of in the tone of "We should do this for our own good." Also in 1992, DuPont stated that sale of C2F6 to semiconductor manufacturers would be curtailed if suitable emissions controls were not in place by a set point in the future. C2F6 was targeted because it was (and still is) the most widely used PFC gas in the chamber clean process. When DuPont issued its first statement in 1992, the technology to reduce PFC emissions from the semiconductor industry was immature at best. When introduced in the 1980s, PFCs were seen as a welcome substitute from the ozone-depleting substances that were being phased out. Desirable properties included price, ease of use and low toxicity.
|
|
The 1996 Memorandum of Understanding (MOU) between the Environmental Protection Agency and the semiconductor industry, a cooperative effort between these two parties to reduce PFC emissions, has been prominent in effecting action. Analogous partnerships throughout the world (Japan, Korea, Taiwan, Europe) with similar goals also have been forged, all leading to the WSC target of 10% reduction by 2010. A brief but concise summary of the global emissions reduction efforts can be found elsewhere.10
PFC emission reduction strategies
Driven initially by DuPont's 1992 statement, the U.S. semiconductor industry has taken a four-pronged approach to reducing PFC emissions. These include process optimization, substitution of non-global warming gases, recovery/reclaim of unused PFCs, and point of use (POU) abatement. The focus has been on both CVD chamber cleans (60-70% of total emissions) and etch processes. All four methods have been pursued aggressively since the mid-1990s with some impressive results.
Optimization: It had been shown that the default CVD chamber clean recipes from the tool manufacturers were not optimized. Through the use of endpoint monitoring of effluent gases during the chamber cleans, the use of C2F6 was shown to be in excess by as much as 25% on certain processes.11 As C2F6 is the primary chamber-cleaning PFC gas in terms of bulk consumption, optimization is a quick way to reduce emissions by a substantial amount. Unfortunately, the initial investment in time is notable, as optimization is dependent both on process and tool type. Despite an initially significant amount of effort with endpoint monitoring, chamber clean optimization also proves to be a significant source of savings, especially for larger fabs.
Conversely, etch processes typically are optimized for the proper ratio of C, O, and F atoms to preserve anisotropy, and therefore are less available for emissions reduction through optimization efforts. However, information on Lam's new Excelan etch chamber describes up to 90% PFC emissions reduction compared to the company's older Rainbow 4520 oxide etch reactors.12 These reductions were due to careful optimization of recipes as well as significant hardware and chamber design contributions.
Substitution: Perhaps the easiest approach would be that taken by the layman: "Don't use global warming gases." Alternatives to PFC gases undoubtedly will be significant in the ultimate answer to this issue. Unfortunately, PFCs have properties that make them difficult to replace. Much work has been done on hydrofluorocarbons (HFCs), unsaturated PFCs and iodo compounds, just to name a few.13,14 These compounds are marked with significantly shorter atmospheric lifetimes, as well as lower GWP values in most cases. However, most either prove too expensive for chamber cleans or tend to be unsuitable for etch due to anisotropy, polymerization or other factors.
C3F8 has been promoted as a drop-in replacement for C2F6. C3F8 exhibits a shorter atmospheric lifetime (2,600 years, Table 3) vs. C2F6, as well as a substantially higher utilization in the plasma (about 55%, vs. 30% for C2F6 and 10%-15% for CF4).15 Though currently somewhat more expensive than C2F6, C3F8 is used in lower quantities, so it is economically comparable. Lower consumption results in lower MMTCE emission values, but the conversion of C3F8 to CF4 in the plasma is significant.16
Chlorine trifluoride (ClF3) is used for the chamber clean step in some processes. ClF3 is significantly reactive to effect a chamber clean without plasma — the clean is strictly thermal. Chamber cleans are more complete, as the reactive region is not restricted to the plasma field — corners are cleaned and particulates reduced. Proponents of ClF3 point out that in addition to its chamber cleaning capabilities, it is a non-global-warming gas (atmospheric lifetime is measured in minutes). However, its high reactivity increases health and safety concerns, as a leak or other mishap with ClF3 would have the potential for a significant hazard to equipment and personnel. Proper precautions and abatement of ClF3 are imperative.
Perhaps the most effective substitution of non-global warmers to date is the use of NF3. IBM developed the dilute NF3 chamber clean recipe in which NF3 and helium are used in place of C2F6.17 Since the conversion of NF3 is significantly higher in the plasma than C2F6, the total flow of NF3 is lower, and MMTCE emissions are reduced to 5-10% based on C2F6 emissions for a similar process. Novellus has successfully incorporated this technology in some of its newer tool sets.18 NF3 also figures prominently in Applied Materials' Remote Chamber Clean.19 This method involves decomposition of NF3 to fluorine gas (F2) in a microwave discharge chamber directly adjacent to the wafer chamber.20 The F2 (and presumably atomic fluorine) perform the chamber clean in the plasma-free environment of the chamber. The NF3 is converted to F2 with >99% efficiency, significantly reducing the global warming potential of the effluents. This chamber clean process will be standard on certain future Applied Materials tools, and is retrofittable to selected tools. As with ClF3, F2 is so reactive that it has no global warming potential. Also in common with ClF3, proper abatement of F2 is imperative. Entrainment of F2 into a dry bed system, while efficient, is expensive, as the reactive beds are consumed rapidly by the high flows of chamber clean gases. Conversion of F2 to HF in a point-of-use abatement device and subsequent water scrubbing is generally recommended to minimize HF (and particulates, TEOS, etc.) in the ductwork. Concentration of the HF in the wastewater through recirculation in the scrubber allows this process stream to be treated efficiently in the fluoride treatment facility.
Recovery/recycle: Considered the "green" approach, recovery and reclamation involves capture of unused PFC gases for disposal or purification and future use. Recovery efforts originally were targeted primarily at C2F6, as this gas was the predominant PFC gas for chamber cleans (over 80%) at the time. Targeting of C2F6 also makes sense from an efficiency point of view; only about 25% of the C2F6 in a chamber clean is actually utilized; the rest is pumped away and would be available for reclaim. Separation of the C2F6 from the CF4, O2, HF, SiF4, N2 and other gases present in the exhaust stream has been the main technological hurdle. Beyond that, DuPont has said it would accept recovered C2F6 for purification or disposal as appropriate.
As a consequence, the major gas companies (BOC, Air Products, Air Liquide and Praxair) have all demonstrated feasible technologies to recover PFCs from the exhaust stream. Air Products21 and Air Liquide22 have had success with membrane separations, and Praxair has a cold-box approach for removing PFCs from the N2 and O2.23 Several commercial systems have been developed. In each of these systems, the exhaust gas required extensive pretreatment to remove acid gases, particulates and water before exposure to the membrane or cold separation platform. The required pretreatment is comparable in cost to POU abatement of PFCs. BOC Edwards proved the viability of PFC capture from the vacuum pump before N2 dilution, but no commercial product was developed.24 Despite the excellent technological work on PFC reclaim/recovery, the price of C2F6 is currently too low to make recovery an economically viable option. Conversion to electronic grade also is problematic. Recovered C2F6 would have different impurities than DuPont currently gets from its virgin feedstock, making purification batch-dependent. It has been estimated that purification would add about $6/kg of recovered C2F6 above and beyond transportation, repackaging, etc. Off-site abatement probably could be performed for $3/kg.25 Cost considerations probably would dictate that the recovered PFCs be destroyed instead.
Abatement: PFC abatement is perhaps the most mature of the four technologies discussed. However, it also had been considered the least desirable approach to emissions reduction as it has relatively little up-front value added. Only recently has it become apparent that abatement will be an integral part of the overall solution. This turnaround is attributed partly to the fact that etch emissions have proven very difficult to reduce using optimization and substitution. Etch emissions vary from CVD chamber clean emissions by an order of magnitude or more. While abatement has been available for CVD chamber cleans flows(>1 slm of CF4 or C2F6) since 1997, this technology is viewed as overkill for etch, for which PFC flows are much smaller. To address these issues, abatement systems targeted at etch processes recently have been developed.
BOC Edwards developed the Thermal Processor Unit and demonstrated its capacity to destroy PFCs. This combustor/scrubber combination has even demonstrated the capacity to remove high flows of CF4.26,27 However, this technology initially proved too expensive for widespread approval. Recent developments have dramatically reduced its cost of ownership, including 90% reduction of water consumption.28 While a combustor that uses natural gas as fuel will generate CO2, the resulting reduction in global warming potential can be greater than 99.5% when compared to the GWP of the PFCs abated. Similarly, HAP (hazardous air pollutant) emissions, such as HF and NOx, are very low.
Other abatement technologies recently have been commercialized. A catalytic PFC abatement system has been introduced by Hitachi that is targeted at etch processes. This system features a wet scrubber for particulate and acid gas removal, followed by a heated catalyst bed to convert the PFCs to HF and CO2. A second scrubber removes the HF formed.29 The second scrubber greatly reduces HAPs emissions, and the catalytic bed operates for months while reducing PFC emissions from etch processes.
| TABLE 4. Cosignatoriesto Memorandum ofUnderstanding betweenthe U.S. EPA and theSemiconductor Industry |
| Advanced Micro Devices |
| American Microsystems Inc. (AMI) |
| Burr-Brown Corp. |
| Cherry Semiconductor Corp. |
| Dominion Semiconductor LLC |
| Eastman Kodak Co. |
| Hewlett-Packard Co. |
| Intel Corp. |
| International Business Machines Corp. |
| LSI Logic Corp. |
| Lucent Technologies, Microelectronics Group |
| Micron Technology Inc. |
| Motorola Inc. |
| National Security Agency |
| National Semiconductor Corp. |
| NEC Electronics Inc. |
| Philips Electronics North America Corp. |
| Rockwell Semiconductor Systems |
| ST Microelectronics Inc. |
| Sony Semiconductor Co. of America |
| Texas Instruments Inc. |
Another new technology involves destruction of the PFC gases in the exhaust foreline through the use of Litmas Blue's plasma discharge apparatus.30 In the foreline between the turbo and dry pumps, a plasma discharge is produced, which decomposes the PFCs into component carbon and fluorine atoms. To prevent recombination into CF4 and related compounds, a source of hydrogen atoms (typically H2, water vapor or CH4) is added to the foreline in approximately stoichiometric quantities to ensure conversion to HF. Tests show the %DRE of PFCs is greater than 90%, although conversion does drop at higher PFC flow rates. There seems to be no corrosion on the vacuum pump by HF, but removal of HF from the pump exhaust may require post-pump abatement.
Summary
The emission of perfluorocompounds from the semiconductor industry has been targeted for reduction using a multi-pronged approach. The implication of global warming by PFCs and related compounds has been reviewed in both technical and political forums. While overall emissions of global warming gases by the semiconductor industry are relatively low, the industry has moved with swiftness to address the issue. Through intense research and a worldwide effort, several technologies — optimization, substitution, recovery and abatement — have been developed that should enable semiconductor manufacturers to achieve the World Semiconductor Council goal of 10% reduction by 2010 based on 1995 values. Recent developments have significantly reduced the cost of abatement methodologies to the point where abatement is now economically feasible on a large scale. •
Dr. Joe Van Gompel is a product specialist for exhaust management systems at BOC Edwards. He is based in Austin, Texas, and does customer and sales support for the point-of-use exhaust management devices. He has a Ph.D. in organic chemistry from the University of Illinois, and has worked as a bench chemist and also as a customer support/applications specialist for a major FTIR manufacturer. He is active in global warming issues as they relate to PFCs and the semiconductor industry. He has a patent and several publications, and he is a member of the American Chemical Society and the Society of Applied Spectroscopy.REFERENCES
- W. Worth, "Total Perfluorocompound (PFC) Emissions Data Report," Sematech Report 96073156B-ENG, Dec. 31, 1997.
- See http://www.ipcc.ch/pub/techrep.htm
- "Inventory of US Greenhouse Gas Emissions and Sinks," US Environmental Protection Agency, 1990-1998.
- D. Cowles, "Oxide Etch Tool Emissions Comparison for C5F8 and C4F8 Process Recipes," A Partnership for PFC Emissions Reductions, SEMICON Southwest, Austin, Texas, Oct. 18, 1999.
- International Technology Roadmap for Semiconductors, 1999 Edition, p. 248.
- See www.eeca.org/pdf/pfc_emis.pdf
- L. Beu, W. Worth, "A Review of PFC Emissions Reporting Methodologies and Suggestions for Improvements," A Partnership for PFC Emissions Reductions, Semicon Southwest, Austin, Texas, Oct. 13, 1997.
- See www.idg.net/crd_council_76214.html
- Dataquest, "Year end 1998 Forecast; Capital Sending in the Wafer Fab Equipment and Silicon Markets," Report SEMM-WW-MT-9901, March 8, 1999.
- L. Beu, P.T. Brown, "An Analysis of Fluorinated Compound Emissions Reduction Technologies and Emission Reduction Goals," Semiconductor Fabtech, 11th Edition (Spring 2000), p. 91.
- J. McNabb, "C2F6 Recipe Optimization at Hewlett Packard," A Partnership for PFC Emissions Reductions, Semicon Southwest, Austin, Texas, Oct. 13, 1997.
- A. Doe, J. Maille, G. Hills, R. Ridgeway, T. Strencosky, "Empirical Modeling of Emissions from Dielectric Etch Reactors for the Assessment of their Environmental Impact," Environmental Impact of Process Tools, SEMICON West, July 12, 1999.
- Y. Yoshida, Japan's "Research on Alternative Etching Gases," Global Semiconductor Industry Conference on Perfluorocompound Emissions Control, Monterey, Calif., April 7, 1998.
- S. Karecki, L. Pruette, R. Reif, "Development of Alternative Chemistries for Wafer Patterning and PECVD Chamber Cleaning Applications," Global Semiconductor Industry Conference on Perfluorocompound Emissions Control, Monterey, Calif., April 7, 1998.
- W. Worth, "Tool Perfluorocompound (PFC) Emissions Data Report," Sematech Technology Transfer 96073156B-ENG, Dec. 31, 1997.
- A. Cheng, M. Daniels, "Evaluation of C3F8 and TFAA as Chamber Cleaning Gases in an Applied Materials 5200 TEOS Deposition Tool," Sematech Report 98063537A-ENG, July 31, 1998.
- T. Tamayo, "IBM's Success with Process Optimization and Beyond," Global Semiconductor Industry Conference on Perfluorocompound Emissions Control, Monterey, Calif., April 8, 1998.
- K. Aitchison, "Simultaneous Resolution of Multiple Environmental Issues: The Development of In-Situ RF Plasma Chamber Clean Processes Using Dilute NF3He Mixtures," Semiconductor Fabtech, 11th Edition (Spring 2000), p. 87.
- L. Mendicino, S. Filipiak, P.T. Brown, J. Langan, R. Ridgeway, A. Johnson, R. Pearce, P. Maroulis, A. Atherton, "Evaluation of the Applied Materials µClean Technology for DxZ Chamber Clean for Perfluorocompound (PFC) Emissions Reduction," Sematech Report 98083547A-TR, Aug. 31, 1998.
- L. Mendicino, P.T. Brown, S. Filipiak, L. Beu, A. Johnson, R. Pearce, P. Maroulis, R. Basnett, W. Holber, "Motorola Evaluation of the Applied Science and Technology Inc. (ASTeX) ASTRON Technology for Perfluorocompound (PFC) Emissions Reductions on the Applied Materials DxL Chemical Vapor Deposition (CVD) Chamber," Sematech Report 99033697A-TR, April 16, 1999.
- W. Carson, K. Christian, E. Crossland, T. Hsiung, R. Ridgeway, J. Yang, "Large Scale PFC Caputure System, A Partnership for PFC Emissions Reduction," SEMICON Southwest, Austin, Texas, Oct. 13, 1997.
- W. Cummins, T. Gilliland, S. Kesari, D. Miner, G. Fleming, K. Trilli, "The Future of Perfluorocarbon Capture and Recycling: Membrane Technology," Semiconductor International, Vol 20, #8, July 1997, p. 265.
- T. Gilliland, C. Hoover, "Evaluation of Praxair's Perfluorocompound (PFC) Capture/Recovery System," Sematech Report 98113600A-ENG, Dec. 21, 1998.
- T. Gilliland, A. Seeley, "Edwards Perfluorocompound Recovery at Texas Instruments," A Partnership for PFC Emissions Reduction, SEMICON Southwest, Austin, Texas, Oct. 13, 1997.
- M. Mocella, "PFC Recovery: Issues, Technologies and Considerations for Post-Recovery Processing," Global Semiconductor Industry Conference on Perfluorocompound Emissions Control, Monterey, Calif., April 7, 1998.
- T. Walling, A. Tran, R. Ridgeway, "Evaluation of an Edwards TPU 4214 and an Ecosys Phoenix IV for CF4 Abatement (ESJC002)," Sematech Technology Transfer 97073319A-TR, Sept. 30, 1997.
- J. Van Gompel, T. Walling, "A New Way to Treat Process Exhaust to Remove CF4," Semiconductor International, Vol 20 (#10), September 1997, p. 95.
- J. Van Gompel, V. Chidgopkar, "Reducing Water Use in Exhaust Management Systems," Micro, Vol 17 (#7), July/August 1999, p. 67.
- S. Tamata, "Catalytic Destruction of PFCs," Global Semiconductor Industry Conference on Perfluorocompound Emissions Control, Monterey, Calif., April 8, 1998.
- V. Vartanian, L. Beu, T. Stephens, J. Rivers, B. Perez, E. Tonnis, M. Kiehlbauch, D. Graves, "Long-Term Evaluation of the Litmas Blue' Plasma Device for Point-of-Use (POU) Perfluorocompound and Hydrofluorocarbon Abatement," Sematech Report 99123865A-ENG, Jan. 7, 2000.
