Harmful Compounds Yield to Nonthermal Plasma Reactor
New gaseous emissions abatement method can be tailored to treat specific industrial effluents
It is well known that airborne emissions from the manufacturing, transportation, and utility industries can contribute to ozone-induced smog and acid rain, as well as potential global warming and atmospheric ozone depletion mechanisms. Increasingly stringent environmental regulations and policies intended to reduce air pollution have necessitated the development of alternative abatement methods for a wide variety of harmful gaseous compounds. An industry's failure to comply with the provisions and amendments of the Clean Air Act of 1990, the Federal Facilities Compliance Act, and other state, federal, and international air standards can result in costly fines and interruption of operations.
Compounds subject to regulation include volatile organic compounds (VOC) such as toluene and methylene chloride, perfluorinated compounds (PFC) such as nitrogen trifluoride and sulfur hexafluoride, oxides of nitrogen (NOx) and sulphur (SOx), and particulate matter (PM).
In the microelectronics industry, for example, PFCs are used extensively for such processes as plasma enhanced chemical vapor deposition, plasma etching, and plasma cleaning. Applications such as semiconductor wafer cleaning and oxidation enhancement employ a wide variety of VOCs, many of which have been identified as ozone depleters or associated with potential global warming mechanisms.
In general, only a small portion of harmful gases is consumed during semiconductor fabrication, so the effluent stream from a particular process may contain relatively large quantities of toxic and environmentally harmful compounds. While the ideal solution to this problem would be to develop cleaner processing methods, or to recover and recycle unused process chemicals, achieving the zero emissions goal sought by the microelectronics industry probably will require continued use of some form of aftertreatment.
As another example, consider that 39 percent of NOx released into the atmosphere originates from mobile sources. Of the many environmental laws in force, those covering NOx and particulate emissions from diesel engines are among the most sweeping. Although significant emissions reductions have been achieved through improved engine design and electronic control, engineering trade-offs may make further simultaneous reductions of NOx and particulates impractical, because engine modifications designed to reduce NOx emissions usually result in increased particulate emissions, and vice versa. Faced with this challenge, engine manufacturers are considering novel forms of exhaust aftertreatment to meet tougher engine emissions standards.
The effective destruction of harmful gases is further complicated by the fact that such emissions are not composed of a single pollutant compound entrained in a simple background gas, but rather a complex mixture of compounds, making the precise chemical composition of a wastestream difficult to characterize. Therefore, a treatment option that is highly generic, or able to destroy multiple pollutant compounds, is desired.
In collaboration with industry and the military, Southwest Research Institute (SwRI) engineers are developing a new technology that transforms environmentally harmful air pollutants into less harmful constituents. Called the pulsed corona reactor (PCR), the technology consists of a nonthermal plasma reaction process under atmospheric pressure. Tests conducted at SwRI have shown that the PCR has the potential to neutralize a wide variety of pollutants in an energy-efficient manner.
Conventional Air Pollution Treatment Options
Conventional emissions control methods rely on four different technologies: thermal, wet, dry, and subatmospheric plasma. Two or more of these techniques are sometimes combined in a treatment scheme; for instance, a semiconductor manufacturing facility may employ thermal, wet, and subatmospheric plasma methods to achieve high levels of pollutant destruction. Each approach has distinct advantages but also suffers from serious limitations in light of new, more stringent regulations and operational requirements.
Thermal incineration entails mixing the process exhaust with hydrogen or oxygen and then passing the mixture through an ignitor where the effluent gases burn. This technique, in which the entire wastestream is elevated to a high temperature to treat a dilute pollutant concentration (typically a few hundred parts per million), is routinely employed in semiconductor fabrication and other manufacturing facilities. Thus, process energy goes not only into destroying unwanted compounds, but also into heating the vastly more abundant background gas. In addition to these energy-efficiency concerns, the high operating temperatures subject surrounding structures to thermal stress, which can lead to exhaustive equipment maintenance schedules and the generation of unwanted reaction by-products. Also, the amount of oxygen or hydrogen added is usually substantial, adding to operating costs.
Wet reactors operate by passing the exhaust gas through an absorbing spray of water or other chemicals. Limited solubility of some pollutants in the scrubbing liquid is a concern, as is eventual disposal of pollutant-saturated liquids. This type of reactor is the final stage of an overall hazardous gas abatement system at many industrial facilities.
In dry reactors, the exhaust gases pass through a bed of granules held at either room temperature or above. The room temperature condition results in chemisorption, while higher temperatures cause the pollutant to decompose and react with the granules. A combination of thermal and dry techniques can be found in a catalyst bed, which is used to lower overall activation energy for the chemical reaction of interest. Catalysts beds are widely used in industry; however, their effectiveness tends to be limited to a single pollutant compound; they are inherently non-generic. In addition, they are susceptible to poisoning, a condition in which undesired variables cause the reaction to cease prematurely.
Subatmospheric (0.1-10 Torr) plasma reactors are used to destroy hazardous gaseous compounds emanating from industrial facilities requiring vacuum conditions to maintain cleanliness and precise process control. While under vacuum, exhaust gases from the process chamber pass through a nonthermal plasma, where they are broken down. A disadvantage of this technique is that solid-phase by-products can accumulate on the reactor's internal surfaces or back-diffuse toward the processing reactor, corrupting the operating environment, or be swept from the reactor into the vacuum pump, causing damage to the pumping mechanism.
An innovative alternative to existing techniques that promises to be both highly generic and energy-efficient is a method based on an atmospheric pressure, nonthermal plasma reaction. The following describes the operational theories and characteristics of nonthermal plasma reactors, as well as the results of tests conducted during a project involving the Institute's Automation and Data Systems Division and Automotive Products and Emissions Research Division, and researchers from the Naval Surface Warfare Center and Air Products and Chemicals, Inc. Included in this discussion is a unique application of the technology that has proven capable of simultaneously removing NOx and PM in diesel exhaust emissions.
Plasma -- the Fourth State of Matter
A plasma can be described as a gas to which a specific amount of energy has been added to separate the gas component molecules into a collection of ions, electrons, charge-neutral gas molecules, and other species in varying degrees of excitation. Depending on the amount of energy added, the resulting plasma can be characterized as thermal or nonthermal.
In a thermal plasma, enough energy is introduced so the plasma constituents are in thermal equilibrium -- the ions and electrons are, on average, at the same temperature. An electrical arc is one example of a thermal plasma, a familiar manifestation of which is a lightning bolt bridging the gap between a storm cloud and the earth. The temperature of thermal plasma components is about 1-2 electron-volts (1 eV is associated with 11,600 K).
A nonthermal plasma is one in which the mean electron energy, or temperature, is considerably higher than that of the bulk-gas molecules. Because energy is added to the electrons instead of the ions and background gas molecules, the electrons can attain energies of from 1-10 eV, while the background gas remains at ambient temperature. This nonthermal condition can be created at both atmospheric and subatmospheric pressures, but high volume throughput, and therefore high treatment rates, can only occur at atmospheric pressures.
Because the electrons are preferentially excited in a nonthermal plasma, leaving the more massive ions with lower energy, a significant energy savings can be realized. This energy-savings potential is a primary reason why nonthermal technologies are being investigated at SwRI.
Generating Nonthermal Plasmas
Techniques employed in the generation of atmospheric-pressure, nonthermal plasmas can be categorized as either electron-beam or electrical discharge-based.
In an electron-beam system, electrons are created and then accelerated to high energies under vacuum conditions in a vessel outside the intended reaction chamber, and then introduced into the reaction chamber through a thin metal or semiconducting window. The electrons bombard the background gas, causing a plasma to form. The principal problem with this approach is development of a suitable window. Not only must the window be thin enough to allow the accelerated electrons to penetrate into the reaction chamber, but it must also be robust enough to maintain the vacuum/atmospheric interface and withstand the physical stresses created by heat resulting from the beam. To effectively penetrate conventional metal-foil windows, an acceleration voltage of more than 150 kV can be required. As an added complication, electrons of this energy can lead to the formation of X-rays, which requires appropriate shielding. Recently, ultra-thin semiconductor windows have been developed that may reduce the required acceleration voltage to 75 kV or less; however, their electrical and physical performance remains to be verified.
Discharge-based systems have the advantage of producing high-energy electrons directly within the treated gas volume through the application of locally intense electric fields. Electrons created in or entering these regions can be accelerated to high energies and, in turn, can collide with molecules in the background gas, freeing more electrons in the process. In this way, self-propagating electron avalanches, called streamers or coronas, can be formed within the gas volume. Once created, streamers must be terminated quickly through external or self-quenching mechanisms before the much heavier ions left behind in the streamer channel gain enough energy from the imposed electric field to transition to the arc or thermal plasma condition.
The PCR is a discharge-based device in which a series of high-voltage pulses is applied to a thin metal wire located coaxially within a metal tube. Applying pulses to this geometry generates a number of streamers within the gas volume that emanate from the wire to the cylinder. The faster the pulse, the more streamers per unit length are produced. Similarly, the higher the pulse repetition frequency, the greater the volume of gas that can be treated per unit of time. The width of the applied electrical pulses is minimized to prevent undesirable streamer thermalization. For most geometries, this requires pulses with risetimes of a few nanoseconds (ns) and pulsewidths of less than 100 ns.
Once generated, the plasma electrons collide with the background gas molecules, creating chemically active species known as radicals. The radicals can preferentially react with pollutant molecules in the gas stream, breaking them down into less hazardous or more easily handled compounds. As in a catalyst, reactions normally constrained to occur at high temperatures are now possible at essentially room temperatures. A representative chemical reaction in a PCR is illustrated in the sidebar titled, " Plasma Chemistry in the Pulsed Corona Reactor."
Semiconductor Industry Emissions
In cooperation with the Naval Surface Warfare Center and Air Products and Chemicals, SwRI researchers investigated the PCR as an abatement technique for the treatment of semiconductor processing tool effluent. The reactor used in these tests consisted of 10 one-inch diameter, three-foot long reaction chambers operating in parallel. Extensive, fast electrical diagnostics were used to accurately determine the electrical efficiency of the reactor. Stable operation of the reactor was demonstrated at voltages up to 35,000 V and repetition rates to 1,500 pulses per second (Hz). The risetime of the pulses was 6 ns, with an overall pulsewidth of less than 100 ns.
The chemical and electrical performance of the PCR was analyzed for the following pollutant compounds: toluene (C7H8), methylene chloride (CH2Cl2), 1,1,1-trichloroethane (CH3CCl3), dichlorodifluoromethane (CCl2F2), nitrogen trifluoride (NF3), hexafluoroethane (C2F6), carbon tetrafluoride (CF4), and sulfur hexafluoride (SF6).
The background gas containing the pollutant to be destroyed plays a significant role in that destruction. Although some applications specifically require the exclusive use of ambient air, many industrial processes are not so constrained. For instance, diesel exhaust is essentially a lean air mixture, whereas the effluent from a semiconductor fabrication facility is primarily nitrogen. Furthermore, the addition of small concentrations of certain reagent gases to the process stream was found to be beneficial in enhancing the destruction efficiencies of some compounds, as illustrated in the figure previously shown. Additives such as hydrogen or oxygen can act as recombination sites for dissociated components of the pollutant compound. For example, the use of a hydrogen additive dramatically improved the destruction efficiency attainable for NF3 by allowing the dissociated fluorine atoms to be removed in the form of hydrofluoric acid (HF).
With the reactor bounded by 100-300 watts of input power, process-stream flow rates of 5-10 liters per minute in air or nitrogen, pollutant concentrations of a few hundred ppm, and with hydrogen or oxygen as additives (a few hundred ppm), significant removal levels of certain VOCs and PFCs were achieved. Typical energy densities applied to the reactor were less than 1,000 Joules/liter (the energy density applied to the gaseous wastestream for a particular destruction efficiency forms a basis of comparison with other nonthermal plasma reaction schemes). The VOCs C7H8, CH2Cl2, and CH3CCl3, as well as NF3, were effectively removed from the gas stream, with destruction levels exceeding 99.9 percent. Eighty five percent of the CCl2F2 was destroyed. The greater than 99.9 percent destruction figure indicates that the detection limits of the gas chromatograph or mass spectrometer (MS) were surpassed.
Remarkable levels of destruction also were obtained for the extremely stable PFC molecules: 32 percent in the case of C2F6 (with 50 percent O2) and CF4 (with 8 percent O2), and 72 percent in the case of SF6. Thermal methods require a substantially higher energy expenditure to achieve similar chemical destruction efficiencies. Further improvement of nonthermal chemical destruction efficiencies is being pursued through variation of the additive species, manipulation of additive concentrations, and optimization of PCR electrical operating parameters.
Having shown significant removal for the compounds of interest, the next logical step was to identify the reaction by-products. For each process gas of interest, the chemical content of the post-reactor effluent was determined by taking a series of MS scans. Static, or time-integrated, scans were taken of the background gas, the background gas with the PCR running, the background gas plus pollutant and/or additive, and, finally, the full process gas stream with the PCR running. The scans were compared to identify the components showing the most significant change, and these were subsequently monitored during an MS dynamic scan, which measured compound concentration over time, with the PCR running and idle. The figure at left above shows the results of a dynamic scan of the principal by-products, HF and fluorine gas (F2), obtained during destruction of NF3 in a background of nitrogen containing a hydrogen additive. As shown, the decrease in NF3 concentration shows up as a corresponding increase in the concentration of HF and F2. Both HF and F2 are water soluble and therefore easily removed by a wet reactor, and subsequently neutralized or recycled. The original pollutant compound, NF3, could not have been removed by conventional techniques without expending significant amounts of energy. For the compounds of interest to the semiconductor industry, the reaction by-products identified to this point appear to be easily removed with a simple wet reactor.
Diesel Exhaust NOx/Particulate Removal
No single commercially available aftertreatment technique for diesel exhaust has the combined effect of reducing particulate emissions while also affecting gaseous emissions, specifically NOx. Another type of nonthermal plasma reactor, the packed-bed reactor, has been investigated at SwRI as a means of obtaining simultaneous particulate and NOx reduction in diesel engine exhaust.[8,9] The packed-bed reactor contains a multitude of dielectric pellets contained between a set of high-voltage electrodes. The reactor operates on essentially the same principles as the PCR, except the streamer discharges take place in the gaseous voids existing between individual pellets.
The performance of a reactor bed provided by AEA Technology in Great Britain was evaluated on two different light-duty diesel vehicles representing two different engine technologies -- an older indirect-injection Toyota truck and a newer direct-injection Dodge truck.
Exhaust flow rates from 1-7 liters/ minute through a small-scale reactor were examined to determine the effect of gas flow rate on the operation of the plasma aftertreatment system. The results showed that simultaneous removal of NOx and PM in diesel exhaust is indeed possible with a discharge-based plasma reactor, although the efficacy of the technique predictably declines at higher flow rates for a given device geometry. An increase in carbon monoxide (CO) was detected at the reactor exit, with the CO concentration falling off at higher space velocities. Because concentrations of CO in diesel exhaust are typically low, this level of undesired by-product may be acceptable. The performance of the reactor was also evaluated over a light-duty transient cycle with similar results.
Although the underlying concept of nonthermal plasma destruction of a wide variety of pollutant molecules has been demonstrated, establishing the viability of the nonthermal plasma approach will require not only examination of its chemical efficiency under a wide variety of conditions, but also measurement of electrical efficiency and prime power requirements, identification of all reaction by-products, and demonstration of the devices' reliability, maintainability, and scalability.
In view of these concerns, a consortium of diesel engine and equipment manufacturers has been formed to further investigate various methods of nonthermal plasma generation for the reduction of NOx and particulate emissions. The most promising devices will undergo additional development and testing to determine the practical limits and operating characteristics of these devices in a realistic diesel exhaust aftertreatment scenario. A similar approach is being pursued to determine the impact of nonthermal plasma systems on multiconstituent semiconductor processing tool effluent.
Published in the Spring 1996 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.