Has the Challenge of PFCs Really Been Solved?
In the semiconductor industry, chemical vapor deposition (CVD) chambers were traditionally cleaned using perfluorocarbon (PFC) gases such as C2F6 and CF4.
Michael R. Czerniak and Kirel Tang, Edwards, Wilmington, Mass.; Shou-Nan Li, Industrial Technology Research Institute (ITRI), Hsinchu, Taiwan -- Semiconductor International, 10/1/2007
In the semiconductor industry, chemical vapor deposition (CVD) chambers were traditionally cleaned using perfluorocarbon (PFC) gases such as C2F6 and CF4. These gases decomposed in a plasma within the process chamber, yielding highly reactive fluorine radicals that reacted with process deposits on the chamber walls and furniture. The volatile fluorinated gases thus formed (e.g., SiF4) could then be pumped away. A relatively small fraction (typically ~30%) of the cleaning gas actually took part in the chamber cleaning reactions; the remainder simply passed through into the exhaust — and the atmosphere. The global warming effects of C2F6 and CF4 are well documented.1,2
In 1999, the World Semiconductor Council (WSC) established a global industry goal of reducing absolute PFC emissions by 10% relative to the 1995 baseline by the year 2010.3 Considering that industry growth between 1995 and the declaration in 1999 was already significant, this declaration represented a dramatic commitment. A major reduction in PFC emissions was quickly achieved by switching from C2F6 and CF4 to NF3, which decomposes very efficiently into fluorine and nitrogen in a plasma, either in or near the process chamber. In effect, this changed the abatement challenge from PFC destruction to fluorine treatment.
Although fluorine does not directly contribute to global warming, it is still a toxic gas that cannot be released indiscriminately into the atmosphere. It is difficult to remove with wet scrubbing alone, and is typically addressed by passing the CVD exhaust stream through a combustion abatement process. Heat supplied by a burning air/fuel mixture thermally reacts various components into byproducts that are easily removed. Silicon-based components from CVD gases, such as silane or TEOS, are oxidized into silica, and fluorine reacts with hydrogen from the fuel to form hydrogen fluoride, both of which can be removed from the wastewater stream in a final washing stage following the combustor. The gas that is finally released to the environment (atmosphere) is essentially free of hazardous and global warming emissions. Figure 1 illustrates the combustion abatement process.
| 1. Schematic representation of a chemical vapor deposition (CVD) exhaust configuration using an open-flame combustor. |
The simplicity of this process is attractive, but a detailed examination of the chemical and thermal processes occurring in an open-flame configuration reveals fundamental problems. The first problem is the de novo generation of PFCs — the very same compounds the industry sought to eliminate with the switch to NF3 — that results from the burning of the hydrocarbon fuel, typically methane, in intimate contact with fluorine from the process chamber (Fig. 2). This reaction is represented below4:
4F2 + CH4 → CF4 + 4HF
The literature includes numerous observations of PFC creation in open-flame combustion processes.5
The second problem is the formation of NOx, which results from the poorly controlled thermal environment of an open flame configuration. NOx is a collective term for several highly reactive gases6 composed of nitrogen and oxygen in various amounts. Many of the nitrogen oxides are colorless and odorless, although one common pollutant, nitrogen dioxide (NO2), frequently combines with particles in the air to form a hazy, reddish-brown layer overlying urban areas. NOx is among the main ingredients involved in the formation of ground-level ozone, which can trigger serious respiratory problems. It reacts to form nitrate particles and acid aerosols, which also cause respiratory problems, and it contributes to the formation of acid rain, to nutrient overloads that deteriorate water quality, and to atmospheric particles that impair visibility. NOx reacts to form toxic chemicals, and contributes to global warming. Because NOx and the pollutants formed from it are long-lived and can be transported over great distances, successful control strategies must be regional, or even global, in scope. NOx emissions are closely regulated in many territories around the world.
Research shows that NOx emissions do not form in significant amounts until flame temperatures reach ~1500°C (Fig. 3).7 Once that threshold is passed, any further rise in temperature causes a rapid increase in the rate of NOx formation. Thermal NOx is produced during the combustion process when nitrogen and oxygen are present at elevated temperatures as they combine to form NO and NO2.
The formation of NOx in open-flame combustors has also been reported extensively (Fig. 4).5
Inward-fired combustion
An alternative to the open-flame configuration, known as an inward-fired combustor, minimizes the formation of both PFCs and NOx in the abatement process. In this design, the fuel and air mixture enters the chamber and combusts on the surface of a cylindrical porous ceramic pad. Process gases are introduced independently through inlets at the top of the combustion chamber. The distinguishing feature of this configuration is that the combustion of fuel, which provides the necessary heat, and the thermal oxidation/reduction of the process exhaust gases are spatially separated, eliminating any opportunity for unwanted cross-reactions between the hydrocarbon fuel and fluorine. In the inward-fired design, the fluorine encounters only water and carbon dioxide, the products of the combustion reaction. In subsequent reactions, the dynamics strongly favor the formation of hydrogen fluoride over perfluorocarbons, as the fluorine combines more readily with the hydrogen from water molecules than with the carbon now tightly bound up in carbon dioxide.
CH4 + 2O2 → 2H2O + CO2
2F2 + 2H2O → 4HF + O2
The inward-fired combustor design effectively prevents the creation of PFCs in the combustion process (Fig. 5). The efficiency of this design at avoiding the unwanted cross-reaction between fuel and fluorine (from NF3) is demonstrated in Figure 6, where CF4 is below detectable limits, even for significant NF3 flows.
| 5. Schematic diagram showing how the inward-firing combustor avoids forming CF |
The inward-fired combustor also provides much better control of thermal conditions throughout the combustion chamber. The reaction volume is completely enclosed, and heat radiates uniformly inward from the cylindrical ceramic pad, creating isothermal conditions throughout. Exhaust gases are fully constrained to pass through the heated zone with all paths exposed to essentially equivalent thermal conditions throughout their transit. This contrasts dramatically with open-flame designs, where different paths experience quite different conditions, and even the best paths may include only brief segments at optimal temperature.
| 6. Comparison of NF3 gas input (green) into an inward-firing combustor to the CF4 output gas (blue), clearly illustrating that there is no generation of CF4 in the combustor. |
The inward-fired configuration is both more effective and more efficient. By ensuring that all gas reaches the correct reaction temperature, it ensures effective and complete abatement. By physically confining the gas within the combustor, it permits higher flow rates through a smaller burner. By effectively capturing and distributing the heat of combustion within the isothermal region, it improves the fuel efficiency of the abatement process. Finally, the constant flow of the fuel/air mixture inward through the combustion pad prevents the accumulation of solid residues, improving overall throughput by reducing maintenance requirements and associated downtime.
The thermal uniformity provided by the inward-fired configuration avoids the formation of NOx compounds. In an open-flame configuration, operating conditions that yield the best overall abatement effectiveness and energy efficiency will necessarily include regions within the reaction volume where temperatures are greater or less than ideal. Where the temperature is too low, abatement is incomplete; where it is too high (>1500°C), NOx can form. Such high temperatures are completely avoided in the isothermal conditions of the inward-firing combustor, where the combustion zone is typically maintained at a temperature <800°C — sufficient to completely oxidize silane and TEOS byproducts and efficiently react the fluorine to hydrogen fluoride, but well below the temperatures required to form significant quantities of NOx.
Other aspects of the inward-fired combustor also reduce the opportunity for NOx formation. The physical separation of combusting fuel from the exhaust gas precludes many NOx forming reaction pathways that would be enabled in an open-flame configuration by the availability of hydrogen from the methane fuel. Operating the inward-fired combustor at lean fuel conditions further reduces available hydrogen in the reaction volume.8 Finally, the extensive use of ceramics for the combustor pad and other high-temperature components eliminates hot metal surfaces that can catalyze NOx formation reactions. It has been reported that NOx concentrations in the exhaust of an inward-fired combustor were below instrumental detection limits, <1.4 ppm.9 The same report noted that another unwanted byproduct, OF2, was also below detectable limits (<1 ppm).
The switch to NF3-based cleaning processes ultimately resulted in a major shift in cleaning strategies within the industry. With C2F6-based processes, cleaning rates were limited by the relatively low efficiency at which the precursor gas molecules could be converted to fluorine radicals (typically <30%). Current generation plasma sources and the high electron attachment cross-section of NF3 provide very high conversion rates, and engineers have been able to dramatically reduce chamber cleaning times by simply increasing the NF3 flow rates. Recent expansions in NF3 manufacturing capacity and resulting reductions in price have added further incentives to increase NF3 flow to reduce cleaning downtime and increase process throughput and productivity.
With fluorine levels in cleaning exhaust typically <10%, abatement systems must cope with the challenge of effectively treating the fluorine entrained in huge flows of nitrogen. The confined flow, isothermal reaction volume and strong reducing environment of the inward-fired combustor ensure complete conversion of fluorine to hydrogen fluoride, even at high flow rates.
Summary
The use of NF3 reduces PFC gas emissions from CVD chamber cleaning processes, but care must be taken not to generate PFCs in the combustion abatement processes used to remove fluorine from the exhaust stream. All combustors are not the same. Open-flame combustors, which mix the exhaust gas directly with the burning hydrocarbon fuel, cannot prevent unwanted reactions between carbon in the fuel and fluorine in the exhaust that form CF4. In addition, hydrogen from the fuel, high temperatures and the presence of hot metal surfaces catalyse the formation of NOx, an air pollutant. Inward-fired combustion solves both problems by physically separating the burning fuel from the exhaust gas and providing tightly controlled, isothermal conditions throughout the reaction volume. The inward-fired design handles higher exhaust gas flows through smaller combustors, delivers better fuel efficiency, and reduces downtime for maintenance and cleaning — ultimately reducing both the environmental impact and cost of ownership for inward-fired configurations.
| Author Information |
| Michael R. Czerniak is the product marketing manager for exhaust gas management at Edwards. He received his Ph.D. at Manchester University, and started as a scientist at Philips' UK laboratories before moving to its fab in Nijmegen, working on compound semiconductor applications. Phone.: +44 (0)1275 337100 Email Mike.Czerniak@edwardsvacuum.com |
| Kirel Tang is the general manager of Asia Solar & FPD business at Edwards. He received his Ph.D. in physics from the University of Cambridge, UK Email: Kirel.tang@edwardsvacuum.com |
| Shou-Nan Li works at the Center for Environmental, Safety and Health Technology at the Industrial Technology Research Institute (ITRI). He received his Ph.D. from the University of Florida, and is a department manager at Energy & Environment Research Laboratories. Email: SNLi@itri.org.tw |
| References |
| 1. J.V. Gompel, "PFCs in the Semiconductor Industry: A Primer," Semiconductor International, July 2000, p. 321. |
| 2. J.T. Houghton et al., Climate Change 2001: The Scientific Basis: Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change, 2001, Vol. 1, Cambridge University Press. |
| 3. World Semiconductor Council, PFC Emission Press Release, May 2005. |
| 4. D.R. Burgess et al., "Thermochemical and Chemical Kinetic Data for Fluorinated Hydrocarbons," Prog. Energy Combust. Sci., 1996, Vol. 21, p. 453. |
| 5. S.N. Li, J.N. Hsu, S.J. Lin, J.L. Hong and H.Y. Shih, "FTIR Spectrometers Measure Scrubber Abatement Efficiencies," Solid-State Technology, July 2002. |
| 6. U.S. Environmental Protection Agency, "NOx — How Nitrogen Oxides Affect the Way We Live and Breathe," September 1998. |
| 7. Alentec, "The Formation of NOx." Available at www.alentecinc.com/papers/NOx/The%20formation%20of%20NOx_files/The%20formation%20of%20NOx.htm. |
| 8. Z.M. Djurisic, "NOx Formation in Ultralow-NOx Gas Burners," ACERC Conf., 2003. |
| 9. M. Waltrip, D. Baker and M. Czerniak, Evaluation of the BOC Edwards Thermal Conditioning System for HAPs Abatement, SEMICON West 2000. |