Catalytic Process for Control of PFC Emissions

		At a Glance
	Catalytic abatement processes, widely used in the chemical industry, have only recently been applied to the control of PFC emissions. The development of a new PFC abatement process based on catalytic destruction is described.

Perfluorocompounds, examples of which include CF₄, C₂F₆ and SF₆, are used in the manufacture of semiconductor materials. Although the emissions of these materials are small in terms of total mass emitted, the emissions are significant in terms of global warming potential. This is because the atmospheric lifetimes of PFCs are on the order of many thousands of years, resulting in global warming potential (GWP) of many times that of CO₂.

The semiconductor manufacturing industry is operating under a Memorandum of Understanding (MOU) with the Environmental Protection Agency (EPA).¹ Under this MOU, the industry is targeting emissions reductions to 10% below 1995 levels by 2010. Significant emissions reductions have already been achieved through process optimization and alternative chemistries (e.g. adaptation to NF₃). Industry growth models, however, indicate that these reductions will not be sufficient to meet future PFC emissions needs.²

Abatement technologies offer an effective means of meeting current and future PFC emissions reduction schedules. Several abatement technologies have been evaluated in the past few years. Examples include thermal incineration,³ plasma abatement,⁴ and catalytic destruction.⁵

Plasma abatement units are located between the tool exhaust and the dry pump. This results in one plasma abatement unit being required per chamber, thereby presenting no economy of scale. Also, in the event unscheduled maintenance issues arise, the tool will need to be taken off-line so that the plasma abatement unit can be serviced. Because it operates at low pressure, plasma abatement does not lend itself well to integration with SiF₄ and HF scrubbers. Thus, a separate scrubber may be required downstream of the plasma abatement process.

Thermal incineration and catalytic destruction technologies are located downstream of the dry pump and designed to treat emissions from multiple chambers, thereby providing economy of scale. Also, the downstream location of the catalytic and thermal unit will allow the tool to continue operating should maintenance issues arise. These processes also lend themselves well to SiF₄ and HF scrubber integration.

The catalytic destruction process differs from the thermal incineration process in that a catalyst is used to destroy the PFCs. The use of a catalyst offers several advantages over the thermal process. These advantages include lower operating temperature, smaller volume, reduced energy costs, more controlled process, no formation of partial or incomplete combustion products, and no formation of thermal NO_x. In addition, the lower operating temperature allows the catalytic process to be operated with electric power. Therefore, there is no need to install lines to deliver combustion gases (e.g. methane) to the catalytic abatement process.

Catalytic abatement processes are widely used in the chemical process industry to treat volatile organic compound (VOC) emissions. Only recently has this process been applied to the control of PFC emissions.

The catalytic process for controlling PFC emissions involves passing the PFC-laden stream through a catalyst bed at an elevated temperature. It is within the catalyst bed that the PFCs are converted (in the presence of water and air) to CO₂ and HF. The catalyst is the key to the successful implementation of the catalytic abatement technology. For the catalyst to be employed in a commercial PFC abatement process, it must be both highly reactive and durable, i.e. able to maintain a high destruction efficiency for an extended period of time.

Catalyst assessment

The design of a catalytic abatement process required first defining the performance envelope of the catalyst. Defining the performance envelope involves: 1) understanding the overall mechanisms governing the destruction of the PFCs; 2) determining the temperature required to achieve the desired destruction efficiency; and 3) assessing the stability and poison resistance of the catalyst.

Unlike the thermal incineration process, whereby PFCs are decomposed via a reaction with gas phase oxygen, the catalytic decomposition of PFCs proceeds according to a catalyzed hydrolysis reaction. Therefore, water, rather than oxygen, is necessary for the decomposition reaction to proceed. For example, the reactions governing the catalytic decomposition of C₂F₆ are:

C₂F₆ + 3H₂O ® CO + CO₂ + 6HF

Reaction 1

CO + ½O₂ ® CO₂

Reaction 2

Details of the catalyzed hydrolysis reaction have been published elsewhere.⁶ Although not directly involved in the catalytic decomposition of PFCs, oxygen contributes to the reaction scheme by providing a mechanism for the conversion of CO to CO₂ (Reaction 2). Thus, the decomposition of PFCs in humid air will result in CO₂ and HF being the only reaction products formed, whereas the decomposition of PFCs in humid nitrogen will result in the formation of CO.

Light-off curves are typically used to determine the operating temperature of the catalyst within the reactor. A light-off curve is nothing more than a plot showing the destruction efficiency as a function of reaction temperature. The temperature required to achieve the desired level of destruction is selected as the operating temperature for the process. The Table lists the temperature required for achieving >95% destruction (T₉₅) of several fluorine-containing compounds of interest to the semiconductor manufacturing industry. For a given PFC, T₉₅ is a function of the residence time and concentration. Increasing the residence time (i.e. using more catalyst to treat a given process stream) will decrease T₉₅, while increasing the PFC concentration will typically increase T₉₅. Therefore, when using light-off curves to determine the operating temperature of the process, care should be taken to record light-off curves at conditions (concentration and flow rate) consistent with that of the process.

Listing of T95 for Selected Fluorine-Containing Compounds
Compound	T95
NF3	375°C
CHF3	425°C
SF6	510°C
CF4	650°C
C2F6	690°C
C3F8	690°C
c-C4F8	705°C

Results presented in the Table indicate that an operating temperature of about 700°C will be required to achieve >95% destruction of the C₁-C₄ PFCs, and an operating temperature of about 510°C will be required for SF₆. Although reaction temperatures reported in the Table are in many cases significantly greater than temperatures required for the catalytic destruction of VOCs (~200-400°C), the temperature is significantly lower than the temperature required for the thermal incineration of PFCs (1100-1300°C).

Prior to incorporation of a catalyst into an abatement process, the catalyst's stability and poison resistance must be determined. Catalyst stability refers to the ability of the catalyst to maintain a high destruction efficiency for an extended period of time. Catalyst poisoning refers to the decrease in catalytic activity resulting from the presence of impurities (i.e. non-PFC entities) in the process stream. For etch and CVD applications, silicon tetrafluoride (SiF₄) is present in the waste stream and is a potential catalyst poison. This is because silicon-containing compounds are known to deposit onto the surface of the catalyst, physically blocking reactive sites responsible for the destruction of the target compound.

1. CF 4 light-off curves for catalyst prior to and following exposure to CF 4 for 2000 hours, and for catalyst following exposure to CF 4 plus 60 ppm SiF 4 for one week.

Figure 1 reports CF₄ light-off curves for catalyst prior to and following exposure to CF₄ for 2000 hours, and for catalyst following exposure to CF₄ plus 60 ppm SiF₄ for one week. As results show, following 2000 hours of exposure to CF₄, the position of the light-off curve is consistent with that of the fresh catalyst, indicating that the catalyst has retained its reactivity following the extended exposure to CF₄.

Following exposure to the CF₄/SiF₄ mixture, the position of the light-off curve has shifted ~150°C to the right, meaning that higher temperatures are now required to achieve a similar level of destruction (with respect to the fresh catalyst). This result demonstrates that a small amount of SiF₄ will rapidly deactivate the catalyst. Thus, the successful operation of the catalytic PFC abatement process will require that SiF₄ be scrubbed upstream of the reactor.

Catalytic abatement process

Based on results of the catalyst assessment, a PFC abatement unit was designed by Guild Associates. A water pre-scrubber was incorporated into the system design to remove SiF₄ upstream of the catalyst. The size and energy requirements of the system were optimized based on a detailed trade-off analysis that took into account the PFC concentration, flow rate and catalyst life time. A post-scrubber was incorporated into the process to remove the product HF. The post-scrubber was added because the need was recognized to decrease the HF load to the house scrubber and minimize HF transport through the house duct.

2. Schematic representation of the PFC abatement process.

A schematic representation of the PFC abatement process is illustrated in Figure 2. Exhaust gas from up to four chambers of the tool enters the PFC abatement unit, where it is mixed with room air, then delivered to a water pre-scrubber to remove SiF₄ and HF from the chamber exhaust gas. Greater than 99.5% SiF₄ removal efficiency was required to prevent the catalyst from being poisoned.

A Misonix MTS-100 was selected for this pre-scrubbing operation because of its overall efficiency and compact size. The scrubber is composed of four individual scrubbing stages in series. A nitrogen-shielded inlet keeps the inlet section dry and clear and prevents the process gases from hydrolyzing prior to entering the scrubber. This design also keeps the vacuum exhaust line free and clear.

Once inside the pre-scrubber, SiF₄ is first subjected to a hydrolysis chamber, where it reacts with water vapor to form SiO₂ (s) and HF (g). The second stage is composed of a packed section where the acid gases are scrubbed. The stream then passes through a venturi stage, where the majority of the SiO₂ is removed.

A final packed section followed by a high-efficiency mist eliminator ensures that SiF₄ has been completely reacted and subsequently removed from the stream. The Misonix pre-scrubber requires 0.5 gpm of water and occupies <4 ft³. The pre-scrubber unit lent itself well to integration and provided the required water to the reactor feed stream.

Following the water scrubber, the humidified PFC-laden stream enters a heat exchanger, where it is heated by recovering heat associated with the hot reactor exhaust. Use of a heat exchanger greatly reduced the energy requirements of the process. An in-line electric heater is located between the heat exchanger and the catalytic reactor. The electric heater is used to heat the process stream to the target reaction temperature (700°C). The catalytic reactor is next in line.

The reactor consists of a stainless-steel shell filled with catalyst. It is within the catalytic reactor that the PFCs are converted to CO₂ and HF. The hot reactor exhaust is passed through the heat exchanger, where the heat associated with the stream is recovered.

Following the heat exchanger, the reactor effluent stream is delivered to a quench stage and then to the water post-scrubber (Misonix MVS-50). The post-scrubber is designed to remove 99% of the product HF from the reactor effluent by employing engineered structured packing material. The post-scrubber utilizes the wastewater from the pre-scrubber and an additional 0.5 gpm of water. Scrubbing HF at the reactor exhaust greatly reduces the load on the house scrubber.

PFC abatement test results

A PFC abatement unit was tested at Applied Materials Inc. (Sunnyvale, Calif.). The system was operated in conjunction with a four-chamber dielectric etch tool. The purpose of the testing was to assess DREs for different PFCs, catalyst stability and process reliability.

About 60,000 wafers were processed over the course of the five-month evaluation. During this time, the system achieved DREs of >99.5% for CF₄, 99% for C₂F₆ and 97% for c-C₄F₈. No decrease in the catalytic activity was observed during the evaluation. This result indicated that the catalyst was stable and that the Misonix pre-scrubber effectively filtered SiF₄ from the tool exhaust. Only CO₂ and HF were detected in the reactor effluent stream. No F₂, OF₂, CO, or products of incomplete oxidation were detected.

The SiF₄ filtration efficiency of the pre-scrubber was determined to be >99.8%, and the HF filtration efficiency of the effluent post-scrubber was determined to be 99%. No system-related downtime was encountered over the course of the evaluation, and the system successfully performed several heat-up/cool-down cycles without incident.

3. Catalyst stability for fresh catalyst and for catalyst removed from the reactor following five months of operation.

Following completion of the evaluation at Applied Materials, the reactor package was removed and shipped to Guild Associates's facility so that the used catalyst could be evaluated. Figure 3 compares results of an accelerated aging test for fresh catalyst and for catalyst removed from the reactor following the five months of evaluation.

The accelerated aging test is designed to deactivate the catalyst in a much shorter time period than normal operation. Following five months of operation, the catalyst is able to maintain >80% of its initial lifetime. Results of this test indicate that the actual lifetime of the catalyst may be in excess of two years. A portion of the used catalyst removed from the reactor was evaluated and determined not to be a hazardous material. Therefore, no special handling of the used catalyst is required.

4. Estimated capital and operating costs on a per-tool basis for multi-tool PFC abatement applications.

Although the catalytic abatement unit as reported in this paper was designed for point-of-use abatement, the process readily lends itself for scale-up to multi-tool applications. Capital and operating costs for multi-tool applications are shown in Figure 4. The data presented include both the cost of abating the PFCs and of scrubbing SiF₄ and HF. When used in multi-tool applications, the capital costs decrease significantly on a per-etch tool basis. By treating emissions from 10 etch tools simultaneously, the capital cost per tool becomes <$40,000 (<$10,000 per chamber of a four-chamber, 200 mm etch tool). Designing the process to treat emissions from several tools negates the need to replicate controllers, valves, duct work, reactor components, etc., resulting in significant cost savings. This decrease in cost illustrates the economy of scale offered by the catalytic abatement process.

Roy S. Brown, executive vice president and COO of Guild Associates, has been involved with the development and design of adsorption and catalyst-based systems for a wide range of applications, including N₂/CH₄ separation systems and catalyst-based pollution abatement systems. He received his B.S. in chemical engineering from California Polytechnic Institute in 1981.
Phone: 1-614-760-8001

Joseph A. Rossin, head of catalyst applications at Guild, has worked almost exclusively in the area of catalysis and reaction engineering. His work has led to the development and commercialization of environmental catalyst for several unique applications, including ammonia, hydrogen cyanide, NO_x and fluorocarbon abatement. He received his Ph.D. in chemical engineering from Virginia Polytechnic Institute and State University (Blacksburg, Va.) in 1986.
Phone: 1-614-760-8007

Christopher J. Thomas, vice president of Misonix Inc. and general manager of the Mystaire division, has more than 12 years' experience in air pollution control for the semiconductor market. His specialty is in point-of-use abatement utilizing aqueous chemistry. He received his B.S. in biochemistry from Villanova University (Villanova, Penn.).
Phone: 1-800-645-9846

---

REFERENCES

1. S. Rand, "MOU-2 Update," SEMICON Southwest, October 2000.
   
2. C. Fraust, "World Semiconductor Council ESH Task Force Update," SEMICON Southwest, October 1999.
   
3. T. Gilliland, "PFC Abatement Technologies: Thermal Oxidation," Global SEMICON Ind. Conf. PFC Emission Cont., 1998.
   
4. V. Vartanian, et al, "Plasma Abatement Reduces PFC Emission," Semiconductor International, June 2000.
   
5. A. Bjatnagar, T. Kaushal, R. Brown, J. Rossin, "Catalytic Abatement of PFC Emissions," SEMICON Southwest, October 1999.
   
6. W.B. Feaver, J.A. Rossin, "The Catalytic Decomposition of CHF₃ over ZrO₂-SO₄," Catalysis Today, Vol. 55, 1999, p. 13.
   

Acknowledgements

The authors wish to thank Ashish Bhatnagar, Tony Kaushal and Sam Shamouilian of Applied Materials Inc. for testing the PFC abatement unit, and the National Science Foundation for support of the catalyst development effort.

Talkback

» SUBMIT FEEDBACK

Good artical. I am interested the ....
keli Yang - 12/08/08

» MORE

Related Content

• Topics
• Author

Related Content

 

By This Author

There are no other articles written by this author.

SPONSORED LINKS