Final Report: Recovery of Perfluorocarbons (PFCs) from Chemical Vapor Deposition Operations in the Semiconductor IndustryEPA Contract Number: 68D70051
Title: Recovery of Perfluorocarbons (PFCs) from Chemical Vapor Deposition Operations in the Semiconductor Industry
Investigators: Wijmans, J. (Hans) G.
Small Business: Membrane Technology and Research Inc.
Project Period: September 1, 1997 through September 1, 1999
Project Amount: $225,000
RFA: Small Business Innovation Research (SBIR) - Phase II (1997) Recipients Lists
Research Category: Small Business Innovation Research (SBIR)
Description:Perfluorocarbons, particularly perfluoroethane (C2F6), are used in the semiconductor industry in plasma cleaning of chemical vapor deposition (CVD) chambers. Although C2F6 is released to the atmosphere in relatively small quantities, it is a very powerful greenhouse gas. The estimated lifetime of C2F6 in the atmosphere is at least 10,000 years, and its global warming potential (GWP) is 13,500 times higher than that of carbon dioxide. In the mid-1990s the semiconductor industry and the U.S. Environmental Protection Agency formed a voluntary partnership to reduce the emissions of C2F6 and other greenhouse gases. The options taken into consideration were C2F6 recovery, C2F6 destruction, and use of C2F6 substitutes with lower environmental impact. The overall objective of this project was to develop an innovative membrane separation process to recover at least 95% of the C2F6 used in CVD operations.
In the CVD chamber cleaning cycle, a mixture of C2F6 in nitrogen is fed to the chamber, where a plasma is applied under high vacuum. The efficiency of the current process is rather low; only 10 to 30% of the C2F6 dissociates into highly reactive ionic species (principally C2F5+). After a cleaning cycle, the gas mixture is pumped from the chamber using a nitrogen-driven vacuum pump. The exhaust stream, containing about 0.5 vol% C2F6 in nitrogen is fed to a water scrubber to remove the ionic species generated during the plasma process. The remaining gas is vented to the atmosphere. A typical semiconductor facility emits about 10,000 to 50,000 lb of C2F6 per year from CVD cleaning processes.
One possible method of reducing the emissions of C2F6 to the atmosphere is to incinerate the off-gas from the cleaning process. However, the operating costs of incineration systems are very high, and a scrubber must also be used to treat the acids in the incinerator effluent gases. The estimated operating cost of a 35,000-lb/year C2F6 treatment plant is $3.7 million/year (including burners and fuel) or $105/lb C2F6. Considering that C2F6 costs $30/lb, this is a significant expense. These high treatment costs led many semiconductor companies to vent the off-gas to the atmosphere.
The proposed membrane process to recover this lost C2F6 consists of two subsystems in series: (1) a nitrogen/C2F6 separation system, which increases the C2F6 concentration in the off-gas, and (2) a C2F6 liquefaction system, which recovers the C2F6 as a liquid. The C2F6 concentration in the stream entering the liquefaction system must be raised to 70-80% C2F6 for condensation to be achieved at reasonable temperatures. The process requires a membrane that will separate nitrogen from C2F6 by preferentially permeating the nitrogen.
The work focused first on identifying a membrane material with high nitrogen permeabilities and intermediate nitrogen/C2F6 selectivities. After evaluating the properties of a number of polymers in the form of composite membranes, the best membranes were scaled up to bench-scale modules and tested in the laboratory. An economic analysis of a C2F6 recovery process based on the experimental data was performed.
Summary/Accomplishments (Outputs/Outcomes):Based on the requirements of the process, eight potential membrane materials from which to prepare high-performance thin-film composite membranes were evaluated. Two membrane materials met the necessary performance target, showing a nitrogen/C2F6 selectivity of at least 200 and a pressure-normalized nitrogen flux of at least 20 H 10-6 cm3(STP)/cm2@s@cmHg. Spiral-wound modules were prepared and tested with mixtures of C2F6 in nitrogen. The very low permeability of the membrane for C2F6 causes the module separation performance to be very sensitive to small defects in either the membrane or the module. The application also requires modules to be operated at very high stage-cuts, making good flow distribution in the feed channels of the module important. By addressing these issues during membrane and module preparation, an effective nitrogen/C2F6 selectivity of 50 was achieved at stage-cuts over 90%. Optimized bench-scale modules were evaluated in parametric tests in the laboratory.
An economic analysis based on the experimental data showed that membrane-based C2F6 recovery is very attractive economically as well as solving an environmental problem. However, the rationale for C2F6 recovery in the semiconductor industry disappeared during 1999 as the industry moved away from using C2F6 as the CVD chamber cleaning gas. This meant that C2F6 recovery in the semiconductor industry was no longer a potential application; therefore, the planned field test of the technology did not take place.
Conclusions:High-performance thin-film composite membranes were successfully developed for the separation of perfluorocarbons from nitrogen. An economic analysis of a C2F6 recovery system based on the properties of these membranes showed that the process was attractive both economically and environmentally. However, the rationale for C2F6 recovery disappeared during 1999 as the industry moved away from using C2F6 as the chamber cleaning gas. The replacements for C2F6 are other perfluorinated chemicals such as nitrogen fluoride (NF3) and perfluorocyclobutane (c-C4F8), which are almost completely consumed during use, leading to minimal emissions to the atmosphere and a reduced global warming impact. Although this means that C2F6 recovery in the semiconductor industry is no longer a potential application, we believe that the membranes developed in this project may find other applications involving perfluorocarbons, for example, to treat process streams generated during their manufacture.
As described above, C2F6 recovery in the semiconductor industry is no longer a potential commercial application. However, the membranes developed in this project are likely to be applied to perfluorocarbon recovery in other situations and for related separations in the petrochemical industry.