Issue



Catalytic technology for PFC emissions control


07/01/2001







Roy S. Brown, Joseph A. Rossin, Guild Associates Inc., Dublin, Ohio
Kenneth Aitchison, Novellus Systems Inc., San Jose, California

overview
Catalytic technology offers a low-cost means of controlling PFC emissions without impacting the semiconductor-manufacturing process. Greater than 99% destruction of C1 to C4 PFCs can be achieved at temperatures of 650-700°C, while PFCs such as NF3 and SF6 can be destroyed at significantly lower temperatures using a novel PFC abatement catalyst. Laboratory and field studies demonstrate that the catalyst is stable, with the ability to maintain high destruction efficiency for an extended period of time. Although stable, the presence of SiF4 in the feed stream poisons the catalyst, necessitating that it be scrubbed upstream of the catalyst bed. Cost of ownership analysis indicates that a catalytic PFC abatement process can treat PFC emissions from etch and CVD tools for $0.20-0.25/wafer.

Perfluorocompounds (PFCs) are some of the most chemically inert substances known [1, 2]. Examples of these compounds include CF4 and C2F6, which are employed during dry chemical etching and chamber cleaning operations in the manufacture of semiconductor devices. Emissions of PFCs by semiconductor-manufacturing processes are under scrutiny due to their contribution to global warming. This contribution, although small in terms of emitted mass, can be significant because of the extremely long atmospheric lifetimes of these compounds. For example, the atmospheric lifetime of CF4 is estimated to be greater than 10,000 years [1].

The primary uses of PFCs in the semiconductor-manufacturing industry are for patterned wafer etching and CVD chamber-cleaning operations. The need for PFC abatement during chamber-cleaning operations has been greatly reduced due to a switch to NF3. NF3 abatement is not required, since greater than 99% of the NF3 is destroyed in the plasma. However, recent economic analyses suggest that it is more economical to use perfluorocompounds in CVD chamber clean operations and abate the unused PFCs than it is to operate the process with NF3.

This is because of: 1) the high relative cost of NF3 versus C2F6 and C3F8; 2) the availability of NF3 on the global market; and 3) the formation of significant amounts of F2 and the economics associated with F2 filtration.

Catalysts are used to control a number of industrial emissions. The catalytic process involves passing the contaminated stream through a catalyst bed at an elevated temperature, typically 150-400°C for volatile organic compound (VOC) abatement. It is within the catalyst bed that the contaminants are reduced to CO2 and H2O, as well as mineral acids if halogens are associated with the parent compound. Although widely used to treat VOC emissions, catalytic technology has only recently been applied to the control of PFC emissions. This article describes the use of catalytic abatement technology to treat PFC emissions related to the manufacture of semiconductors.

Catalytic destruction of PFCs
A catalyst is a substance that facilitates a chemical reaction without being consumed by the reaction. The use of a catalyst lowers the temperature required for a reaction to occur, thereby reducing energy costs. Pollution abatement catalysts are employed commercially to control the emissions of a wide range of organic pollutants. A typical pollution abatement catalyst consists of one or more catalytically active metals (e.g., platinum and palladium) dispersed onto a high surface area substrate (e.g., aluminum oxide), commonly referred to as a "support." This configuration allows for excellent utilization of the catalytic metals, resulting in high reaction rates.

Click here to enlarge image

Figure 1. Light-off curves for CF4, C2F6, C3F8, c-C4F8, CHF3, SF6, and NF3.

The catalyst can be used in several geometric forms, including granules, beads, and extrudates. Alternatively, the catalyst can be coated onto the channels of a monolith. A monolith, or "honeycomb" as it is sometimes described, consists of a series of parallel, noninterconnecting square channels. A thin layer of catalyst support material, termed "washcoat" by the trade, is coated onto the walls of the channel. It is within the washcoat that the catalytic metals are deposited. The monolith offers several advantages over other configurations. These advantages include modular design, low pressure drop, excellent utilization of the catalytic metals, predictable flow characteristics, excellent mass and heat transfer properties, low thermal mass, and low attrition. Because of these advantages, the monolith is often the preferred geometric form.

Catalysts are used in abatement applications for several reasons. First, the use of a catalyst lowers the temperature required for the reaction to proceed, often by hundreds of degrees Celsius. The reduced operating temperature greatly minimizes the energy costs associated with the application. Second, because of the fast reaction rates, catalytic systems are generally smaller than corresponding thermal oxidizers. Contact times needed for catalytic systems are on the order of 0.1-0.3 sec, compared to contact times of 1-2 sec or more for corresponding thermal oxidizers. The shorter contact time translates to a smaller footprint and correspondingly lower capital cost. A third advantage of a catalytic abatement unit is reaction product selectivity. The lower operating temperature associated with the catalytic process eliminates the formation of NOx from the oxidation of molecular nitrogen, as well as products of partial oxidation.

Reactivity of fluorine-containing compounds
"Light-off" curves are used to determine the temperature range over which a compound can be effectively destroyed, and to compare the reactivities of different compounds. Examples of light-off curves for several fluorine-containing compounds of interest to the semiconductor-manufacturing industry are reported in Fig. 1. A light-off curve reports the conversion of a compound as a function of the reaction temperature. Light-off curves are recorded by heating the catalyst to an elevated temperature, introducing the catalyst to the reactant stream, then decreasing the catalyst temperature at a known rate while measuring the concentration of reactant in the effluent stream. The shape and position of a light-off curve is a function of the flow rate of the process stream and the concentration of the reactant compound. Typically, increasing the flow rate of the process stream and/or increasing the concentration of the reactant chemical will shift the light-off curve to the right, meaning that a greater temperature is now required to achieve a similar level of conversion.

Light-off curves corresponding to CF4, C2F6, C3F8, c-C4F8, CHF3, SF6, and NF3 are shown in Fig. 1. Greater than 99% destruction of all of these compounds can be achieved using the catalyst, with NF3 and CHF3 being much more readily destroyed than the C1-C4 perfluorocarbons. The temperatures required for the catalytic destruction of the C1-C4 PFCs (650-700°C) are significantly greater than those associated with traditional VOC abatement applications (150-400°C) [3]. This is attributed to the stability of these compounds that results from the strength of the carbon-fluorine bond.

Temperatures associated with the catalytic destruction of PFCs are, however, significantly less than those associated with the thermal oxidation of PFCs [4, 5]. For example, Banks et al. [4] has reported that temperatures in excess of 1000°C are required to thermally decompose C2F6 and C3F8, and temperatures on the order of 2000°C are required to thermally decompose CF4. Interestingly, CF4 is more readily destroyed by the catalytic process than are C2F6, C3F8, and c-C4F8. This trend is directly opposite to that reported for thermal oxidizers and strongly suggests that the two processes proceed according to different mechanisms.

Click here to enlarge image

Figure 2. Light-off curves for the destruction of CF4 in different environments.

Reaction products associated with the destruction of the C1-C4 perfluorocarbons and CHF3 consist of CO2 and HF. No CO, NOx, or traces of partially destroyed compounds are present. Reaction products associated with the catalytic destruction of NF3 consist of NOx and HF, with the NOx selectivity being close to 100%, i.e., all of the atomic nitrogen associated with NF3 is converted into NOx. NOx emissions resulting from NF3 abatement can be greatly minimized by adding small amounts of ammonia to the process stream. Reaction products associated with the catalytic destruction of SF6 consisted of SOx and HF.

Catalytic implications of PFC reactivity
To be successfully employed in PFC abatement applications, the catalyst must be able to maintain a high level of destruction for an extended period of time (e.g., in excess of one year). The high operating temperatures required to achieve greater than 95% destruction of many PFCs (Fig. 1), coupled with the generation of HF, results in a rather extreme set of operating conditions for the catalyst.

Typical catalysts employed in pollution abatement applications consist of platinum group metals finely dispersed on an aluminum oxide support material. This catalyst is not suitable for PFC abatement applications for a number of reasons. Platinum group metals, although highly reactive oxidization catalysts, can form volatile halide compounds and are thus generally not tolerant of halogens at temperatures greater than about 400°C. Aluminum oxide, although thermally stable to temperatures in excess of 800°C, is readily transformed to AlF3 during the destruction of fluorine-containing compounds [6].

Therefore, nontraditional catalyst formulations are required to treat PFC-laden process streams. Base metals, although not as reactive as platinum group metals, do not readily form halides at elevated temperatures. Nontraditional catalyst support materials, such as TiO2 and ZrO2, are known to retain their integrity in the corrosive HF environment [7, 8], but they do not possess the desired thermal stability. These materials can, however, be stabilized to function at elevated temperatures [9, 10], making them viable components in the design of PFC abatement catalysts.

Reaction mechanism
The primary application for abatement catalysts is to control VOC emissions. The mechanism by which hydrocarbons are oxidized over noble metals involves, as a rate-limiting step, a reaction between adsorbed oxygen and the adsorbed hydrocarbon to yield CO2 and H2O [6]. For this reaction to proceed, it is necessary that oxygen be present in the process stream. As shown in Fig. 2, the catalytic decomposition of CF4 does not require oxygen. Removing oxygen from the feed stream (i.e., conducting the test in N2) has no effect on the shape or position of the CF4 light-off curve. This result demonstrates that the catalytic destruction of CF4 does not proceed according to an oxidation mechanism. Removing water from the feed stream essentially renders the catalyst inactive, allowing one to conclude that water is necessary for the catalytic decomposition of CF4.

The fact that water is required to facilitate the destruction of CF4 indicates that the reaction proceeds according to a catalyzed hydrolysis mechanism. Catalyzed hydrolysis reactions involve an interaction between the reactant chemical (either gas phase or adsorbed) and water adsorbed on the surface of the catalyst, to yield CO2 and HF [11, 12]:

CF4 + 2H2O→ CO2 + 4HF

The actual surface chemistry governing the catalyzed hydrolysis of PFCs has not yet been studied and remains an area for future research.

Catalyst stability and poisoning
The commercial viability of a catalytic process depends on the ability of the catalyst to function for an extended period of time. Because of the high operating temperatures and corrosive HF environment, the stability of a catalyst employed for the destruction of PFCs must be evaluated. Deactivation is the term for a decrease in catalytic activity over time resulting from the decomposition of the target compound.

There are a number of mechanisms that can be responsible for the deactivation of a catalyst [13]. In the case of PFC abatement, deactivation due to fluorine attack and thermal processes must be considered. Deactivation resulting from fluorine attack refers to the transformation of the catalyst, either the support material or catalytic metals, to the fluoride form. For example, Farris et al. [6] attributed the deactivation of a platinum/alumina catalyst to a transformation of the aluminum oxide support to aluminum trifluoride. This transformation resulted in a collapse of the pore structure, preventing access of the reactant chemicals to the catalytically active metals located within the support material. Transformation of the catalytic metals to the fluoride form will alter the corresponding chemical state, potentially rendering the metals inactive.

Deactivation resulting from thermal processes commonly refers to a collapse of the pore structure of the support material brought about by the high operating temperatures. This is often due to a phase change of the support material. For example, high surface area TiO2 (anatase phase) begins to undergo a transformation to the low-surface-area rutile phase at temperatures of approximately 450°C [9]. This phase transformation results in a collapse of the pore structure, minimizing access of the reactant chemicals to catalytic sites within the support, thereby reducing the overall effectiveness of the catalyst.

Click here to enlarge image

Figure 3. Accelerated aging test results for the destruction of C2F6 and SF6 in humid air.

It is often not feasible to evaluate the lifetime of a catalyst under normal operating conditions in the laboratory due to time constraints. The stability (lifetime) of a catalyst can often be accurately evaluated using accelerated aging tests, however.

Such a test is designed to probe the stability of the catalyst on a much shorter time scale. To accomplish this, it is designed to expose the catalyst to more severe conditions than those intended by the application. This usually involves a combination of a greater reactant concentration, shorter contact time, and greater temperature. In this manner, more than a year of operation can be simulated in just a few weeks of testing.

Results of accelerated aging tests conducted with C2F6 and SF6 using the PFC abatement catalyst are shown in Fig. 3. These tests were performed employing a feed concentration of 2000ppm in humid air. In both cases, the feed concentration is on the order of three times greater than that of the intended application, while the contact time is approximately four times shorter. Therefore, the dose of PFCs that the catalyst is exposed to in a year of operation can be achieved in one month. The experiment was also conducted under conditions where the PFC conversion was less than 99%. Maintaining the conversion between 90% and 95% prevents the results from being masked by mass transfer effects, thereby allowing changes in the intrinsic reactivity of the catalyst to be recorded. Over the 1000-hr duration of the tests, no decrease in the conversion of either PFC is observed. These results attest to the durability of the PFC abatement catalyst and demonstrate that the catalyst is able to function for an extended period of time without loss of reactivity.

Click here to enlarge image

Figure 4. Poisoning of PFC abatement catalyst by SiF4 during destruction of CF4.

Poisoning is another mechanism by which a catalyst may be deactivated. Poisoning results from chemical and/or physical interactions between the catalyst and impurities in the process stream. Silicon-containing compounds are well-known catalyst poisons. They act by decomposing on the surface and depositing silicon dioxide within the pore structure of the catalyst. These deposits reduce the catalytic activity by minimizing access of the reactant chemical to the catalytically active sites located within the pore structure.

Silicon tetrafluoride (SiF4) is often present in PFC-laden waste streams. It is possible that moderate concentrations of SiF4 will not affect the PFC abatement catalyst, since any silicon dioxide that deposits on the catalyst may be removed as SiF4 through reaction with HF (generated during the destruction of the PFC). The ability of SiF4 to poison the PFC abatement catalyst is illustrated in Fig. 4. Data presented in this figure correspond to a CF4 concentration of 1000ppm and SiF4 concentrations of 0, 2, and 60ppm in humid air at 650°C.

In the absence of SiF4, the PFC abatement catalyst is stable throughout the duration of the 500-hr test. The addition of just 2ppm SiF4 to the feed stream results in a slow but steady decrease in the catalyst's activity. Increasing the concentration of SiF4 to 60ppm results in a rapid decrease in conversion of CF4. Spectroscopic examination of the catalyst following the run revealed the method by which the catalyst was poisoned. SiF4 poisoned the catalyst by depositing SiO2 at or near the external surface of the catalyst. The deposits of SiO2 blocked the pores of the support material, greatly reducing access of CF4 to catalytic sites located within the support. Results presented in Fig. 4 demonstrate that even small amounts of SiF4 will poison PFC abatement catalysts. Therefore, the successful implementation of catalytic technologies to control PFC emissions will require that SiF4 be scrubbed upstream of the catalyst bed. Fortunately, SiF4 scrubbing technologies already exist and are readily adapted to this technology.

PFC abatement process
A schematic representation of a PFC abatement process is illustrated in Fig. 5. The PFC-laden stream from either a wafer etch or CVD chamber clean process is first passed through a water scrubber to remove SiF4. The water scrubber also serves to humidify the incoming process stream, as required in order to maintain a long catalyst lifetime. The stream is then combined with room air and delivered to a recuperative heat exchanger. The recuperative heat exchanger uses the heat of the hot catalyst bed exhaust stream to heat the incoming process stream. An electric heater follows the recuperative heat exchanger and is used to further heat the process stream to the desired operating temperature (650-700°C). The catalyst bed is located just after the electric heater. It is within the catalyst bed that the PFCs are reduced to CO2 and HF. The hot exhaust stream is passed through the recuperative heat exchanger and exits the system at approximately 150°C. The use of the recuperative heat exchanger to recover heat associated with the hot exhaust stream greatly reduces the energy requirements of the process. The effluent stream is then vented to a post-scrubber that removes the HF generated by the process and further cools the effluent stream.


Figure 5. Process schematic for PFC abatement unit.
Click here to enlarge image

The size and energy requirements of a PFC abatement process will depend upon the application (etch or CVD), the size of wafers being processed by the tool (200 or 300mm), and the number of chambers associated with the tool. For example, a PFC abatement process designed to treat emissions from a four chamber dielectric etch tool will have a footprint of approximately 75 x 80cm and use approximately 0.15m3/min room air and 4.0 liter/min DI water. Because of the heat recovery built into the process, the energy utilization of this system is expected to be between 3-4kW. The system can be located downstream of the vacuum pumps and is not expected to impact the operation of the tool.

Conclusion
A PFC abatement unit employing the catalyst described herein was constructed by Guild Associates Inc. and evaluated [14]. The system was able to achieve >99% destruction of CF4 and C2F6, and >95% destruction of c-C4F8 throughout the duration of the test (>50,000 wafers processed). Cost of ownership is estimated to be $0.20-0.25/wafer, based on utilities cost; installation and site preparation; capital cost; annual catalyst replacement cost; and scheduled PM. Results of this test indicated that the PFC abatement system will provide an integrated solution for treatment of PFCs and hazardous air pollutants for dielectric etch and CVD systems.

Acknowledgments
The authors are indebted to James E. Kotary for his technical assistance. We wish to thank the National Science Foundation for funding the catalyst development portion of this effort.

References

  1. A.R. Ravishankara, S. Solomon, A.A. Turnipseed, R.F. Warren, "Atmospheric Lifetimes of Long-Lived Halogenated Species," Science, Vol. 295, p. 194, 1993.
  2. J. Kiplinger, T. Richmond, C. Osterberg, "Activation of Carbon-Fluorine Bonds by Metal Complexes," Chem. Rev., Vol. 94, p. 373, 1994.
  3. J.J. Spivey, "Complete Catalytic Oxidation of Volatile Organics," Ind. Eng. Chem. Res., Vol. 26, p. 2165, 1987.
  4. R.E. Banks, B.E. Smart, J.C. Tatlow, in Organofluorine Chemistry, Plenum Press, New York, 1994.
  5. T. Gilliland, "PFC Abatement Technologies: Thermal Oxidation," Global Semicon. Ind. Conf. PFC Emission Cont., 1998.
  6. M.M. Farris, A.A. Klinghoffer, J.A. Rossin, D.E. Tevault, "Deactivation of a Pt/Al2O3 Catalyst during the Oxidation of Hexafluoropropylene," Catalysis Today, Vol. 11, p. 501, 1992.
  7. S. Karmakar, H.L. Greene, "An Investigation of CFC12 (CCl2F2) Decomposition on TiO2 Catalyst," J. Catal., Vol. 151, p. 394, 1995.
  8. G.M. Bickle, T. Suzuki, Y. Mitarai, "Catalytic Destruction of Chlorofluorocompounds and Toxic Chlorinated Hydrocarbons," Apld. Catal. B, Vol. 4, p. 141, 1994.
  9. C.A. Leduc, J.M. Campbell, J.A. Rossin, "Effect of Lanthana as a Stabilizing Agent in Titanium Dioxide Support," Ind. Eng. Chem. Res., Vol. 35, p. 2473, 1996.
  10. Manufacturer's literature, Magnesium Electron Inc.
  11. W.B. Feaver, J.A. Rossin, "The Catalytic Decomposition of CHF3 over ZrO2-SO4," Catalysis Today, Vol. 55, p. 13, 1999.
  12. R.S. Brown, W.B. Feaver, J.A. Rossin, "Catalytic Control of Emissions during Semiconductor Manufacture," in Improving Environmental Performance of Wafer Manufacturing Processes, Semicon West, 2000.
  13. C.H. Bartholomew, J.B. Butt, Catalyst Deactivation 1991, Elsevier, New York, 1991.
  14. A. Bhatnagar et al., "Catalytic Abatement of PFC Emissions," in A Partnership for PFC Emissions Reduction, Semicon Southwest 1999.

Roy S. Brown received his BS in chemical engineering from California Polytechnic Institute in 1981. He has worked at Guild Associates since 1982, where he has been involved with the development and design of adsorption- and catalyst-based systems. He is executive VP and COO at Guild.

Joseph A. Rossin received his PhD in chemical engineering from Virginia Polytechnic Institute and State University in 1986. He has worked almost exclusively in the area of catalysis and reaction engineering. Rossin is the head of catalyst applications at Guild Associates Inc., 5750 Shier-Rings Rd., Dublin, OH 43016; ph 614/798-8215, fax 614/798-1972, e-mail [email protected].

Kenneth Aitchison earned his BA in chemistry from New York University and his PhD in chemistry from Queen Mary College, The University of London. He has more than 20 years of experience in the field of materials-processing chemistry and is senior director of environmental, safety, and health systems at Novellus Systems Inc.