Environ. Sci. Technol. 2001, 35, 1587-1592
An Atmospheric-Pressure Plasma Process for C2F6 Removal MOO BEEN CHANG* AND SHENG JEN YU Graduate Institute of Environmental Engineering, National Central University, Chungli 320, Taiwan, R.O.C.
Perfluorocompounds (PFCs) are widely used in the semiconductor industry for plasma etching and chemical vapor deposition (CVD). They are relatively inert gases that intensely absorb infrared radiation and, therefore, aggravate the greenhouse effect. A bench-scale experimental system was designed and constructed to evaluate the effectiveness of C2F6 conversion by using dielectric barrier discharges (DBD) with atmospheric-pressure plasma processing. Experimental results indicated that the removal efficiency of C2F6 increased with applications of higher voltage and frequency. Combined plasma catalysis (CPC) is an innovative way for abatement of PFCs, and experimental results revealed that combining plasma generation with catalysts could effectively enhance C2F6 removal efficiency achieved with DBD. The major products of C2F6 with DBD processing include CO2, COF2, and CO, when O2 was included in the discharge process. Experimental results indicated that as high as 94.5% of C2F6 were removed via CPC at applied voltage of 15 kV, frequency of 240 Hz in the gas stream of N2:Ar:O2:C2F6 ) 50:40:10:0.03.
Introduction Perfluorocompounds (PFCs) such as hexafluoroethane (C2F6) and tetrafluoromethane (CF4) are extensively used as cleaning and etching gases in semiconductor manufacturing processes. These compounds are excellent absorbers of infrared radiation and therefore aggravate global warming. On the other hand, they are also extremely stable gases that are removed very slowly from the atmosphere and have extremely long lifetimes. For instance, the lifetimes of C2F6 and CF4 are 10 000 and 50 000 years, and the GWP100 (global warming potential with time horizon of 100 years) are 9200 and 6500, respectively (1). Great global warming potential and long lifetimes of these gases have triggered a significant effort to develop effective technologies for reducing their release into the environment. In 1997, the Third Conference of the Parties (COP3) was held in Kyoto, Japan; 171 countries developed a treaty that would restrict the release of the most significant greenhouse gases including PFCs. The options evaluated by industry for reducing PFCs emissions include applying substitute chemicals, capture/ recovery/recycle systems, process optimization, and abatement technologies. Developing substitute chemicals that can accomplish the complex process of the semiconductor manufacture and be safely used by factory personnels to replace PFCs is challenging and costly; for instance, the cost * Corresponding author phone: 886-3-4226774; fax: 886-34226774; e-mail:
[email protected]. 10.1021/es001556p CCC: $20.00 Published on Web 03/14/2001
2001 American Chemical Society
of using HFCs (hydrofluorinated compounds) is 5-20 times of the cost of using CF4 and C2F6 (2). ClF3 had been studied for replacing PFCs in the process of semiconductor manufacturing but failed by the generation of undesirable byproducts such as HF and HCl. Membrane separation and cryogenic condensation have the most potential among the options of the capture/recovery/recycle systems, but the cost associated can be significantly greater than that using virgin material. The abatement technologies may include the methods based on combustion, thermal/chemical, conversion, and plasma technology. Combustion is the most developed and commercially available technology (2), but it may create undesirable byproducts (e.g. HF) that need to be scrubbed. The cost of combustion including fuel consumption and scrubbing system is relatively high. The technique of thermal/chemical as catalyst can only treat some PFCs and its applicability is limited by bed capacity. Based on the disadvantages discussed above, the technique of plasma generation has the advantage of high temperature and low power consumption; therefore the technique of plasma for PFCs abatement is of great potential. Nonthermal plasma technologies including surface wave plasma, arc power plasma, and microwave plasma have been investigated in recent years as an alternate technology for controlling PFCs emissions (3-5). However, more equipment is generally needed to operate at low pressure that tends to increase the capital and operating cost. Recently, some researchers started to investigate the PFCs removal via nonthermal plasma technologies under the atmospheric pressure. Yamamoto et al. (6) and Chang et al. (7) had investigated the NF3 decomposition by using ferroelectric packed-bed plasma reactors. Yamamoto et al. (6) used four coaxial ferroelectric packed-bed reactors in series to treat 5000 ppmv NF3 in a N2-O2 mixture gas and achieved a significantly high removal efficiency of 99.6%. A barrier discharge reactor/adsorbent hybrid system was developed by Chang et al. (7) for removing NF3. They found the zeolite adsorbent could effectively enhance the NF3 removal efficiency achieved with barrier discharge, and complete removal of NF3 was possibly attained. Futamura et al. (8) had tested two nonthermal plasma reactors, i.e., ferroelectric packed-bed and silent discharge, for processing CF4; they claimed that CF4 conversion efficiency via silent discharge was a little higher. Urashima et al. (9) evaluated C2F6 removal (1000-3000 ppm) via a ferroelectric packed-bed reactor with parallel electrodes, and the maximum energy efficiency achieved by plasma was about 2.4 g/kWh. This paper describes and demonstrates the concept of applying a gas-phase oxidation process for C2F6 removal. Dielectric barrier discharge (DBD) is applied to generate effective electrons to react with C2F6 to form CFi radicals and O radicals which can further couple together with CFi radicals to convert C2F6 into CO2 and fluorocarbon oxides including COF2. Effective generation of electrons and O radicals is essential for decomposition and removal of C2F6 with this process. The gas-phase oxidation of C2F6 depends on two major mechanisms including direct electrons dissociation and reactions with O radicals generated via DBD. The primary step is to decompose C2F6 by electrons to form CFi (including CF3, CF2, and CF) radicals. CFi radicals can be generated via the collisions between C2F6 and electrons or F radicals which are produced by electron-impact dissociation of C2F6 as shown in reactions 1-7 along with the rate coefficients (k) evaluated at 25 °C. In addition, Ar may enhance the C2F6 dissociation since Ar is much easier to excite (forming Ar*) and ionize (forming Ar+) compared to background gas of N2. Further reactions of Ar* and Ar+ with C2F6 molecules would VOL. 35, NO. 8, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1587
Obviously, increasing the number of electrons will enhance the reaction of primary step as described in reaction 17. Catalytic conversion is a technique based on thermal/ chemical method for PFCs abatement. The process of catalytic decomposition is effective for PFCs destruction because of its lower temperature for decomposition and compactness in space. Combining plasma catalysis (CPC) is an innovative technique developed in this study to remove PFCs. The catalysts that combined with plasma would play the catalysis role to enhance C2F6 conversion.
Experimental Section FIGURE 1. Major pathways leading to C2F6 conversion. enhance C2F6 removal (10-12), as shown in reactions 8 and 9:
The resulting CFi radicals can then react with O radicals and be oxidized to form CO2, COF2, and CO as shown in reactions 10-16 along with the relevant rate coefficients evaluated at 25 °C (10-12).
General scheme for C2F6 decomposition in O2-containing gas is illustrated in Figure 1. Conversion of C2F6 via DBD includes two steps, the primary step is to generate effective electrons to collide with C2F6 to form CFi radicals, and the second step involves with the O radicals reaction with CFi radicals to produce CO2 and fluorocarbon oxides including COF2. The reaction rates of these two steps are listed in reactions 17 and 18.
RI ) kI[e]a[C2F6]b c
RII ) kII[CFi] [O]
d
(17) (18)
(a, b, c, d are the order of reactions; kI, kII are the relevant rate coefficients.) 1588
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 8, 2001
The laboratory-scale experimental system that has been designed and constructed to evaluate the effectiveness of DBD and combined plasma catalysis (CPC) for C2F6 conversion is schematically described in Figure 2. The experimental apparatus was comprised of a continuous-flow gas generation system, laboratory-scale DBD reactor, gases sampling, and detection system. The gas generation system consisted of compressed gas cylinders including N2, O2, C2F6/N2, and Ar and was regulated by a set of mass flow meters (Teledyne Hastimgs-Raydist HFC 202) to control the flow rates of feeding gases. Hence, the composition of the gas streams could be accurately controlled. The gas streams were kept at atmospheric pressure in the DBD reactor, and the inlet flow rates were kept at 0.6 slpm for all tests. The gas residence times within the DBD and CPC systems were different (since the effective volumes were not the same) even if the gas flow rates for these two systems were kept the same. The cylindrical DBD reactor was made of a crystal quartz tube, and the inner electrode was made of a tungsten rod and was aligned along the tube’s centerline. The outer electrode, made of stainless steel mesh, was wrapped around the outside of the crystal quartz tube. Additionally, CuO/ ZnO/Al2O3 catalysts (5.4 mm long and 5.2 mm in diameter pellets) were packed inside the reactor for CPC experiments. The volumes of DBD and CPC reactors were 137 cm3 and 220 cm3, respectively. The gas residence time within the DBD reactor is 12.6 s for the gas flow rate of 0.6 slpm. As for the CPC reactor, the void fraction of the packed catalysts is 0.4, and the gas residence time corresponds to 8.1 s for the gas flow rate of 0.6 slpm. An alternating-current power supply (Chen Hwa, model 2700P) with variable voltage and frequency was applied to the reactor to generate plasma. An oscilloscope (Tektronix, model TDS410) and a 40-kV peak pulse probe with a 1000-times divider (Tektronix, model P6015A) were connected to the reactor to measure the power consumed by the reactor. The power consumption of the whole system were measured by the power meter installed in the power supply. The Fourier transform infrared (FTIR) spectrometer (Bio-Rad, Model FTS 165) was connected online for identification and measurement of C2F6 and end products including CO2, COF2, and CO. Furthermore, the gas chromatograph (GC-TCD and GC-FID, China Chromatography 9800) was used to check the accuracy of the results. The C2F6 conversion efficiency, products’ selectivity, and balance value in the system are defined as follows: C2F6 conversion:
ηC2F6 (%) ×
[C2F6]initial - [C2F6]final [C2F6]initial
× 100%
Product’s selectivity:
S (%) )
[C] × 100% [T]
[C] represents concentration of the product of interest in
FIGURE 2. Schematic of experimental setup.
FIGURE 3. Effect of C2F6 concentration on ηC2F6 in the DBD experiment determined for the gas streams containing 10 vol % O2, 40 vol % Ar, with N2 as the carrier gas. The applied frequency was controlled at 60 Hz. off-gas and [T] represents the summation of total products’ concentrations. Balance Value:
BV (%) ) ×
∑C ∑C
measured
× 100%
feed
∑Cmeasured represents the total moles of carbon (or fluorine) measured in the outlet stream, and ∑Cfeed represents the total moles of carbon (or fluorine) fed into the reactor.
Results and Discussion Dependence of ηC2F6 achieved with DBD on inlet C2F6 concentration ranging from 300 ppmv to 1000 ppmv is shown in Figure 3. The gas stream contained 10 vol % O2, 40 vol % Ar, with N2 as the carrier gas. The applied frequency was controlled at 60 Hz. Experimental results indicate that conversion efficiency of C2F6 increases with increasing applied voltage and decreases with increasing inlet C2F6 concentration. Conversion of C2F6 was mainly achieved with the
generation of energetic electrons and O atoms (as described in reactions 17 and 18). Increasing applied voltage results in the increase of the number of energetic electrons and the internal energy of system. Increasing the number of effective electrons would increase the reaction rate RI as described in reaction 17 and increasing the internal energy of the system would also increase the reaction rate in an endothermic reaction. Conversion efficiency of C2F6 decreased with increasing inlet C2F6 concentration, while the absolute value of C2F6 molecules converted actually increased with increasing inlet C2F6 concentration. Higher inlet C2F6 concentration meant fewer energy available for converting each C2F6 molecule, resulting in lower ηC2F6. The electrons concentration relative to C2F6 concentration decreased with increasing C2F6 concentration in the same applied voltage and experimental results indicated that ηC2F6 decreased with increasing inlet C2F6 concentration. Therefore, the effect of decreasing electrons concentration is greater than the effect of increasing C2F6 concentration to ηC2F6; and a is greater than b in reaction 17 (i.e. RI ) kI[e]a[C2F6]b) for this set of experiment. On the other hand, the existence of Ar in the gas streams will enhance VOL. 35, NO. 8, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1589
FIGURE 4. Dependence of ηC2F6 on applied voltage for select O2 concentration in the DBD experiment. The simulated flue gas stream contained 300 ppmv C2F6, 40 vol % Ar, with N2 as the carrier gas. The applied frequency was controlled at 60 Hz.
FIGURE 5. Dependence of ηC2F6 on frequency for select O2 concentration in the DBD experiment with an applied voltage of 15 kV. The simulated flue gas stream contained 300 ppmv C2F6, 40 vol % Ar, with N2 as the carrier gas. the C2F6 dissociation since Ar is much easier to ionize (forming Ar+) and excite (forming Ar*) compared to background gas of N2, as shown in reactions 8 and 9. The effect of O2 content in the gas stream on ηC2F6 achieved with DBD is also determined (Figure 4). The gas streams contained 300 ppmv C2F6, 40 vol % Ar, with N2 as the carrier gas. For this set of experiments, the frequency was controlled at 60 Hz. Experimental results indicated that the effect of O2 content on ηC2F6 achieved with this process was significant. For instance, ηC2F6 increased from 10% to 34% as the O2 content was increased from 0% to 10 vol % with the applied voltage of 21 kV. Conversion of C2F6 at higher O2 content is possibly attributed to the replacement of fluorine that is bonded on carbon by oxygen as described in reactions 1016. Increasing atomic O density would increase the reaction rate RII in reaction 18 resulting in a higher ηC2F6. However, ηC2F6 started to decrease as the O2 content was further increased from 10% to 20 vol %. Due to the electronegative property of oxygen, too much O2 in the gas stream would actually waste the system energy via O2 dissociation, and this phenomenon inhibited C2F6 molecules from gaining enough energy with an overdosed O2 content, resulting in lower ηC2F6. Dependence of ηC2F6 achieved with DBD on frequency ranging from 60 to 240 Hz for select O2 content is shown in Figure 5. The gas stream contained 300 ppmv C2F6, 40 vol % Ar, with N2 as the carrier gas. The applied voltage was controlled at 15 kV. ηC2F6 increased rapidly as the frequency was increased from 60 to 240 Hz. For instance, ηC2F6 increased from 14.0% to 53.1% as the frequency was increased from 60 to 240 Hz with the O2 content of 10%. Increasing applied frequency represents increasing the number of effective 1590
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 8, 2001
FIGURE 6. Dependence of ηC2F6 on frequency for select O2 concentration in the CPC experiment. The simulated flue gas stream contained 300 ppmv C2F6, 40 vol % Ar, with N2 as the carrier gas. The applied voltage was controlled at 15 kV. electrons and internal energy of the system. Increasing effective electrons would increase reaction rate RI as described in reaction 17. Similar to the effect on increasing reaction rate with increasing applied voltage, increasing the number of effective electrons also increased the collision between electrons and C2F6 molecules, and produced more CF3, C2F5 radicals for conversion with increasing frequency. From the viewpoint of thermodynamics, increasing the internal energy of the system would increase the rate of endothermic reaction, resulting in higher ηC2F6. ηC2F6 achieved with CPC for select frequency and O2 concentration is shown in Figure 6. The gas stream contained 300 ppmv C2F6, 40 vol % Ar, with N2 as the carrier gas. The applied voltage was controlled at 15 kV. The ηC2F6 achieved with CPC is significantly higher than that of DBD. For instance, ηC2F6 was increased from 53.2% for DBD to 94.5% for CPC with a frequency of 240 Hz. Conversion of C2F6 to form CO2 has been demonstrated with the catalysts, and reactions 19 and 20 are the major reactions leading to C2F6 conversion with the catalysts.
(C2F6+ MO f MO-CF3 + CF3, MO-CF3+ 2O f CO2 + F2 + F + MO, where M indicates the catalyst metal atom) (13).
(CF3 + MO f MO-CF2 + F, MO-CF2 + 2O f CO2 + MO + F2, where M indicates the catalyst metal atom) (13). As shown in reaction 19, the first step of catalytic conversion with metals-oxides catalysts involves the C2F6 adsorption on the O atom that is the active site of catalysts, the O atom on catalysts would attack the C atom of the adsorbed C2F6 to form the intermediates and the F3C-CF3 bond is broken at the same time. Thereafter, the O atom of the other metal-oxide adjacent to the intermediate attacks the C of the intermediate, the C-F bond is broken in the meantime, and OdC+-O- are intermediately formed on O; this step is the surface reaction. The O- of OdC+-O- on the metal-oxide attacks the C+ atom of OdC+-O-, and the final step is desorption of CO2 from metals-oxides catalysts.
FIGURE 9. Effect of applied voltage on concentrations and selectivity of products in the CPC experiment. The gas streams contained 300 ppmv C2F6, 10 vol % O2, 40 vol % Ar, with N2 as the carrier gas. The applied frequency was controlled at 60 Hz. FIGURE 7. FTIR spectra at different operating conditions.
FIGURE 8. Effect of applied frequency on concentrations and selectivity of products in the DBD experiment. The gas streams contained 300 ppmv C2F6, 10 vol % O2, 40 vol % Ar, with N2 as the carrier gas. The applied voltage was controlled at 15 kV. Reaction 20 describes other catalytic conversion of CF3 radical that was produced by the first step of reaction 19. The plasma-only reactions such as reactions 1-16 and the plasma combining catalysis reactions such as reactions 19 and 20 both dominated the conversion reaction of CPC; therefore, the catalyst would enhance the conversion efficiency. As with DBD experiments, ηC2F6 achieved with CPC increases with increasing frequency. Increasing applied frequency would increase the number of energetic electrons and the internal energy, resulting in higher ηC2F6 for CPC. FTIR spectra of the plasma conversion of C2F6 are given in Figure 7. The first spectrum shows the components in the gas stream before applying plasma, and the second, third, and fourth spectra show the products of plasma processing with the application of higher applied voltage, applied frequency, and CPC experiment, respectively. Comparing the first spectrum with other spectra indicates that the products include CO2, COF2, and CO, and the unfavorable byproducts may include NO and NO2. When the system is operated at a higher applied voltage, the major product is CO2. CO is detected for the experiment operating at a higher frequency and CPC experiment. However, COF2 does not appear as a product in CPC experiments. The concentrations and selectivity of products achieved with DBD operating at a variable frequency were shown in Figure 8. The gas streams contained 300 ppmv C2F6, 10 vol % O2, 40 vol % Ar, with N2 as the carrier gas. The applied voltage was controlled at 15 kV. The major products formed
in the DBD process included CO2, COF2, and CO that increased with increasing frequency. The internal energy increased with increasing frequency and increasing internal energy would enhance the reaction; therefore, the concentration of products increased with increasing applied frequency. The selectivity of CO2 was the highest of the major products, the products of COF2 and CO would react further with O radicals to form CO2 (COF2 + O f CO2 + F2, CO + O f CO2, respectively), indicating that CO2 was the most favorable product of C2F6 conversion. On the other hand, the number of effective electrons increased with increasing frequency, the products of CO2 and COF2 would further react with electron to form CO by reactions 21 and 22 (e + CO2 f CO + O + e, e + COF2 f CO + F2 + e); therefore, the selectivity of CO increased with increasing frequency. Experimental results also indicated that the carbon balance value ranged from 0.95 to 1.00, and the fluorine balance value ranged from 0.20 to 0.24. The carbons of C2F6 molecule were mainly converted to CO2, COF2, and CO, while the fluorines of C2F6 molecule were converted to COF2, F2 and the other oxide-fluoride such as O2F2 and OF2. A yellow product was found on the reactor wall, and these products might be SiFxOy films deposited during discharging processes. Since these oxide-fluorides are not measured with FTIR and F2 (a homonuclear diatomic gas) cannot be detected by FTIR, oxide-fluorides such as O2F2 and OF2 are the other possible missing compounds for a lower fluorine balance value compared to the carbon balance.
e + CO2 f CO + O + e e + COF2 f CO + F2 + e
6.0 eV
(21)
6.1 eV
(22)
Figure 9 shows the concentrations and selectivity of products in the CPC experiment. The gas streams contained 300 ppmv C2F6, 10 vol % O2, 40 vol % Ar, with N2 as the carrier gas. The frequency was controlled at 60 Hz. The major products of this converting process were CO2 and CO, but no COF2 and other byproducts were detected. The catalysts would play the catalysis role to remove C2F6 within DBD plasma. The major mechanism of C2F6 conversion in CPC experiment as described in reactions 19 and 20 indicated that the product was only of CO2 and that CO would produce by O radicals collision with CO2 molecular (i.e. CO2 + O f O2 + CO) (14). Consequently, the products in CPC experiment would include CO2 and CO. Experimental results also indicated that the carbon balance value ranged from 0.91 to 0.96. The yellow product that observed on DBD experiments did not appear in CPC experiments. VOL. 35, NO. 8, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1591
TABLE 1. Energy Efficiencies for C2F6 Removal via DBD and CPC, Respectively, with the Applied Voltage of 15 kV frequency (Hz)
DBD (g/kWh)
CPC (g/kWh)
60 120 180 240
0.52 0.36 0.29 0.26
1.1 1.7 1.6 0.94
Table 1 presents the energy efficiencies for C2F6 removal achieved with DBD and CPC, respectively. The inlet gas streams contained 300 ppmv C2F6, 10% O2, 40% Ar, with N2 as the carrier gas. The applied voltage was kept at 15 kV, while the frequency varied from 60 to 240 Hz. Results indicate that the energy efficiency of CPC is significantly higher than that achieved with DBD. For instance, the maximum energy efficiency for CPC is approximately 1.7 g/kWh (3130 eV/ molecule) compared with 0.52 g/kWh (9970 eV/molecule) for DBD. Additionally, energy efficiency for CPC increases as the operating frequency is increased from 60 to 120 Hz and starts to decrease as the operating frequency is further increased. On the other hand, the energy efficiency achieved with DBD decreases monotonically with increasing operating frequency. If one compares the energy efficiency in this study with that obtained by Urashima et al. (9), the maximum energy efficiency achieved in this study is 1.7 g/kWh which is around 70% of Urashima et al. (9). Nevertheless, the reactor used by Urashima et al. (9) had a relatively short half-life (60 min or so) due to the carbon deposited on the contact points where discharges most likely occur. This negative effect had not been found in the CPC reactor for more than 6 months’ operation in this study. Further catalytic reactions of these deposited carbons may have prevented this negative effect in the CPC reactor.
1592
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 8, 2001
Literature Cited (1) Houghton, J. T.; Meira Filho, L. G.; Callander, B. A.; Harris, N.; Kattenberg, A.; Maskell, K. Climate Change 1995-The Science of Climate Change, 1st ed.; Cambridge University Press: New York, 1996; p 121. (2) Mocella, M. T. Mater. Res. Soc. Symp. Proc. 1997, 447, 2934. (3) Hartz, C. L.; Ban, J. W.; Jackson, M. W.; Wofforo, B. A. Environ. Sci. Technol. 1998, 32, 682. (4) Wofford, B. A.; Jackson M. W.; Hartz, C.; Bevan, J. W. Environ. Sci. Technol. 1999, 33, 1892. (5) Chen, D. T.; David, M. M.; Tiers, G. V. D.; Schroepfer, J. N. Environ. Sci. Technol. 1998, 32, 3237. (6) Yamamoto, T.; Chang, J. S.; Kostov, K.; Urashima, K.; Okayasu, Y.; Kato, T.; Iwaizumi, T.; Yoshimura, K. J. Adv. Oxid. Technol. 1999, 4, 454. (7) Chang, J. S.; Kostov, K. G.; Urashima, K.; Yamamoto, T.; Okayasu, Y.; Kato, T.; Iwaizumi, T.; Yoshimura, K. IEEE Trans. 1nd. Appl. 2000, 36, 1251. (8) Futamura, S.; Einaga, H. and Zhang, A. IEEE Industry Applications Conference-34th IAS Annual Meeting, Phoenix, Arizona; IEEE Industry Application Society: 1999; p 1105. (9) Urashima, K.; Kostov, K. G.; Chang, J. S.; Yamamoto, T.; Okayasu, Y.; Iwaizumi, T.; Yoshimura, K.; Tato, T. IEEE Industry Applications Conference-34th IAS Annual Meeting, Phoenix, Arizona; IEEE Industry Application Society: 1999; p 1136. (10) Fiala, A.; Kiehlbauch, M.; Mahnovski, S.; Graves, D. B. J. Appl. Phys. 1999, 86, 152. (11) Xu, X. Ph.D. Dissertation, University of Illinois, 1999. (12) Rauf, S.; Kushner, M. J. J. Appl. Phys. 1997, 82, 2805. (13) Gates, B. C. Catalytic Chemistry, Wiley: New York, 1992; p 310. (14) Hsieh, L. T.; Lee, W. J.; Li, C. T.; Chen, C. Y.; Wang, Y. F.; Chang, M. B. J. Chem. Technol. Biotechnol. 1998, 73, 432.
Received for review August 3, 2000. Revised manuscript received January 8, 2001. Accepted January 30, 2001. ES001556P
