Role of methyl nitrate in plasma exhaust treatment

Article abstract of DOI:10.1021/es9813010

There is growing interest in the use of a nonthermal plasma combined with a catalyst for NO_x removal from diesel engine exhaust streams, which contain excess oxygen (6-10%), low concentrations of hydrocarbons (100-1000 ppm), and significant concentrations of water (5-12%). Tests were performed to elucidate the effect of methyl nitrate (CH₃ONO₂) in the plasma-catalyst system since similar compounds have been proposed as reaction intermediates in NO_x removal. NO, CO, CO₂, O₂, SO₂, H₂, Ar, C₃H₆, and C₃H₈ gases were mixed in N₂ carrier. CH₃ONO₂ was prepared and added to the gas blend on test equipment simulating a diesel exhaust gas. CH₃ONO₂ was formed in a substantial yield on passing hydrocarbon/NO_x/O₂/N₂ mixtures through a nonthermal plasma. The catalyst (consisted of two 24 mm x 24 mm long honeycombs) was very effective at removing CH₃ONO₂, which is converted into methanol and NO₂. CH₃ONO₂ was not an important intermediate in the NO_x removal reaction of the plasma-catalyst system.

Full text of DOI:10.1021/es9813010

Environ. Sci. Technol. 1999, 33, 3427-3431
plasma with a variety of catalysts and catalyst substrates,
showed NOx removal from 20 to 80%.
Role of Methyl Nitrate in Plasma
Exhaust Treatment
All of these systems are based on selective catalytic
reaction (SCR) of NOx with a hydrocarbon (HC) reductant.
The HC is present in the exhaust stream due to incomplete
combustion and/ or addition of fuel to the exhaust for the
purpose of increasing NOx reduction efficiency. The mech-
anism of NOx removal may be similar to lean NOx catalysts
operating without a plasma discharge. However, research
such as that of Tonkyn et al. (5) indicates that at least some
catalysts which are effective as conventional catalysts do not
work well with a plasma, and some catalysts which work well
with a plasma do not work without it.
J O H N W . H O A R D , *
T I M O T H Y J . W A L L I N G T O N ,
J A M E S C . B A L L ,
M I C H A E L D . H U R L E Y, A N D
K E N N E T H W O D Z I S Z
Ford Motor Company Research Laboratory,
Dearborn, Michigan 48121-2053
M . L O U B A L M E R
Nonetheless, it is worth noting the reaction schemes
postulated as being significant to nonplasma SCR with HC
reductant (6). It is known that both oxygen and HC must be
present to achieve NOx removal. It is often proposed that the
first step is oxidation of NO to NO2. A catalyst with acidic
surface sites activates the HC in some poorly defined manner.
The rate-limiting step in this process is the pairing of N atoms
to form nitrogen gas (N2). Several steps have been proposed,
including chemisorption of NO or NO2 on the catalyst,
followed by migration of the N atoms along the surface.
Reactions which have been postulated include intermediates
such as nitromethane or methyl nitrate formed by the gas-
phase addition reaction of NO2 to methyl or methoxy radicals
(in the following reactions “M” denotes a third body needed
to remove the excess energy associated with bond formation):
Pacific Northwest National Laboratory,
Richland, Washington 99352
There is growing interest in the use of a nonthermal
plasma combined with a catalyst for NO removal from
x
diesel engine exhaust streams. Such exhaust streams contain
excess oxygen (typically 6-10%), low concentrations of
hydrocarbons (typically 100-1000 ppm), and significant
concentrations of water (typically 5-12%). Conversion of
NO to environmentally acceptable compounds, without
x
requiring a scrubber or an added reductant, is the desired
end result. In our research we observe the formation of
substantial amounts of methyl nitrate (CH ONO ) by the plasma
3
2
CH3 + NO2 + M f CH3NO2 + M
(1)
(2)
discharge. Since similar compounds have been proposed
CH3O + NO2 + M f CH3ONO2 + M
as reaction intermediates in NO removal, tests were
x
performed to elucidate the effect of CH ONO in the plasma-
3
2
catalyst system. CH ONO was prepared and added to
These intermediates may then react with NO2 yielding
The chemistry of a plasma discharge in exhaust gas is
3
2
N2.
the gas blend on test equipment simulating a diesel exhaust
gas. A dielectric barrier plasma discharge was followed
by a zeolite-based catalyst. Methyl nitrate introduced upstream
of the plasma discharge is largely unreacted upon
passing through the plasma. CH ONO arriving at the
complicated. Reference 1 gives a summary of potential
reactions and their underlying kinetics. In the plasma,
electron impact causes dissociation and radical formation;
the radical chemistry creates a mixture of chemical species
in the plasma effluent. Since exhaust has significant amounts
of oxygen and water present, O and OH radicals are common.
Input hydrocarbons react with these radicals, generating a
large number of HC radicals and compounds. While numer-
ous papers have described simplified systems such as NO in
dry air, or NO-H2O-O2 in N2, a detailed description of the
chemistry occurring in a plasma treatment system with a
realistic exhaust gas blend has yet to appear.. A summary of
a system with NO, O2, H2O, and HC, has been reported by
Penetrante et al. (7).
3
2
catalyst is converted to methanol and NO . While methyl
2
nitrate was shown to be formed in this system, it is not a
significant intermediate in the mechanism of conversion
of NO to nitrogen.
x
Introduction
Increasingly stringent air quality standards will require future
diesel engines to be equipped with exhaust aftertreatment
systems capable of decreasing the emission of nitrogen oxides
(NOx). Among systems proposed for diesel NOx reduction
are those based on a nonthermal plasma discharge combined
with a catalyst. Penetrante et al. (1) report testing of a plasma
discharge upstream of a γ-Al2O3 catalyst. Their results indicate
roughly 50% conversion of NOx to N2 in diesel exhaust at 370
°C, 30 J/ L energy deposition, and 18 000 h-1 space velocity
(SV). Hoard and Balmer (2) report that use of a dielectric
barrier discharge plasma followed by a proprietary catalyst
results in roughly 50% NOx conversion at 180 °C and 30 J/ L
energy deposition. Balmer et al. (3) report testing of a plasma
with the same proprietary catalyst used previously (2). The
catalyst was located either in the plasma or downstream of
it. Evidence of N2 formation has been reported during the
operation of the plasma (3). Shimizu and Oda (4), using
The work reported here was undertaken to investigate
the role of methyl nitrate in a plasma based diesel exhaust
treatment system.
Experimental Section
The work was carried out on a flow bench which blends
gases to simulate engine exhaust. Figure 1 is a schematic of
the system. NO, CO, CO2, O2, SO2, H2, Ar, C3H6, and C3H8
gasses are mixed in N2 carrier. Liquid water is injected into
the gas in heated lines. The test plasma and catalyst are in
an oven maintained at 180 °C. The effluent is mixed with
extra N2 to prevent condensation of water at room temper-
ature and passed through measurement instrumentation.
The principal analytical instrument is an FTIR spectrometer
operated at a spectral resolution of 0.125 cm-1 and equipped
with a long path length (20.7 m) sampling cell (8). In addition
* Corresponding author phone: (313) 594-1316; fax: (313) 594-
2923; e-mail: jhoard@ford.com.
9
10.1021/es9813010 CCC: $18.00
Published on Web 08/18/1999
1999 Am erican Chem ical Society
VOL. 33, NO. 19, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 4 2 7

FIGURE 1. Schematic of test lab configuration (FTIR, Fourier transform infrared spectrometer; FID, flame ionization detector for hydrocarbons;
CLA, chemiluminescence NO analyzer).
x
tion with Pacific Northwest National Lab and the U.S.
Department of Energy. Samples of the catalyst can be
TABLE 1. Nominal Input Gas Composition, N Carrier
2
provided free of charge to other investigators for testing upon
gas
nominal concn
gas
nominal concn
request. At 2 L/ min, space velocity (the reciprocal of residence
time) is 4600 h-1. Gas residence time in the oven is
approximately 10 s.
NO
CO
CO2
C3H6
C3H8
250 ppm
400 ppm
7%
1575 ppm (C1)
525 ppm (C1)
Ar
H2
H2O
O2
SO2
1%
133 ppm
5%
8%
90 ppm
The methyl nitrate was prepared by the nitration of
methanol in concentrated sulfuric acid as described by Blatt
(9). The organic phase was dried over CaCl2 and was used
without further purification. FTIR analysis indicated that this
sample has a purity greater than 95%. A small sample (1-2
mL) was transferred to an evacuated metal gas cylinder (size
300) and pressurized to 200 psi with pure nitrogen gas. The
final concentration of CH3ONO2 in the steel cylinder was
measured to be 580 ppm by FTIR. Control experiments were
performed to determine the stability of CH3ONO2/ N2 mix-
tures; there was no observable (<2%) change in the CH3-
ONO2 concentration after standing for 1 week.
Results
CH3ONO2 explodes when heated to its boiling point of 65 °C
(9), so the first question was whether it would in fact survive
the flow through the apparatus consisting of heated lines
and an oven maintained at 180 °C. To answer this question
a control experiment was performed in which mixtures
containing 46 ppm of CH3ONO2 diluted in N2 were flowed
through the system using a total flow of 2 L/ min. The gas
flow was first routed through unheated lines bypassing the
oven to verify flow calibration and then passed through the
heated lines and oven. In both cases no loss of CH3ONO2 was
observed.
Table 2 shows the results from a series of test conditions
with and without CH3ONO2 addition, and with and without
plasma power. These tests were performed with no catalyst
present, so that we are measuring the effect of the plasma
alone. With no added methyl nitrate and no plasma power
(column 3), the measured gas composition matches the
nominal composition; no reactions take place upon passing
through the oven. “HC” is a composite measurement of
several generic hydrocarbon IR bands. Previous work (8)
shows this value correlates well with measurements by an
FID. Also shown is a sum of those measured species which
contain nitrogen:
FIGURE 2. Sketch of the dielectric barrier discharge device.
a Beckman Model OM-11 oxygen analyzer, a Beckman Model
951 chemiluminescent NOx analyzer (CLA) and a Beckman
Model 400 Flame Ionization Detector for HC were also used.
Table 1 lists the nominal composition of the gas blend
used in these experiments. This mixture approximates the
gas composition of diesel engine exhaust, although the HC
concentration is higher than typical modern diesel engines
(typically 100-1000 ppm of C1), simulating introduction of
added HC reductant. Gas flow rate was 2 L/ min and both
plasma and catalyst were held at 180 °C, a typical temperature
for a Partnership for a New Generation of Vehicles (PNGV)-
like vehicle during the Federal Test Procedure.
Figure 2 is a sketch of the dielectric barrier discharge device
used to generate a plasma. It consists of two sheets of alumina,
24 × 100 mm. Each sheet has a gold electrode coated on the
outer side. The two sheets are bonded together with a ceramic
adhesive, leaving a gap of 1.3 mm between the sheets. The
assembly is mounted in a quartz tube such that all the gas
must flow through the gap between sheets. The two electrodes
are connected to a Trek 10/ 10 high-voltage amplifier driven
by an HP33120A function generator. For these tests, the
excitation wave was a 250 Hz triangle wave with peak voltage
between 5 and 6 kV, adjusted so that energy deposition is 30
J/ L (joules of energy per liter of gas flow) when the plasma
is turned on.
N ) [NO] + [NO2] + 2[N2O] + [HONO] +
∑
[CH3ONO2] (3)
Note “a” indicates that the species concentration is below
the detection limit of the FTIR (0.29 ppm for N2O, 10.9 ppm
for HONO, 1.4 ppm for CH3ONO2, 6.12 ppm for CH3OH, and
2.82 ppm for CH2O).
With plasma power on and no methyl nitrate added
(column 4), NO is nearly all converted, primarily to NO2.
Low concentrations of N2O and HONO are generated. At
most a trace of HNO3 is generated, at or below the detection
limit of the FTIR (2.7 ppm). A total of 21 ppm CH3ONO2 is
generated; 43% of the input propene/ propane is consumed,
The catalyst used in these experiments consisted of two
24 mm diameter × 24 mm long honeycombs. These were
made by coating a cordierite ceramic honeycomb (400 cells
per square inch) with a slurry of SiO2 sol gel and catalyst
powder. The catalyst powder is a nonprecious metal, zeolite-
based catalyst, developed under a Low Emission Technologies
Research and Development Partnership project in coopera-
9
3 4 2 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 19, 1999

TABLE 2. Measured Gas Composition with No Catalyst Comparing the Cases of Added CH ONO versus No Added CH ONO , and
3
2
3
2
Plasma Power off versus on
0 methyl nitrate
plasma
46 ppm methyl nitrate
plasma
component
formula
off
on
conversion %
off
on
17
158
4
conversion %
nitric oxide
nitrogen dioxide
nitrous oxide
acids
sum HC
m ethyl nitrate
m ethanol
NO
NO2
N2O
HONO+HNO3
“HC”
CH3ONO2
CH3OH
CH2O
246
9
a
a
9
159
6
3
987
21
96
237
10
1
a
2023
46
93
a
1719
43
20
1353
65
a
115
33
16
a
a
a
15.2
172
a
1
form aldehyde
sum N species
255
204
295
248
a
Below detection lim it.
power-on spectrum, and residual propene was subtracted
from the resulting spectrum of products formed by the
plasma, shown in the third trace from the top. For com-
parison, spectra of methyl nitrate and methanol are also
shown. The presence of CH3ONO2 in the products is clear.
Analysis of the FTIR spectra shows no evidence of other
organic nitrates.
When 46 ppm of CH3ONO2 is added to the flow with power
off (column 6), it is present in the output with no other
significant changes. With plasma power on, the CH3ONO2
concentration increases by roughly the same amount as the
CH3ONO2 formation in the earlier case. This implies that the
added CH3ONO2 survives passage through the plasma without
being reacted. There are no other significant changes in the
concentrations.
Table 3 shows similar data, but with the zeolite-based
catalyst in the flow downstream of the plasma. Without added
CH3ONO2 and with the plasma off (column 3), no conversion
of NOx takes place, and the output gas is very much like the
input gas composition. With plasma on and no CH3ONO2
addition (column 4), there is significant conversion of NO
and 35% reduction in total N-containing species measured.
Prior work (2, 3) suggests that with this catalyst the NOx
conversion is largely to N2, unlike the noncatalyst case. HC
conversion, formaldehyde production, and methanol pro-
duction are all similar to the noncatalyst case. Unlike the
noncatalyst case, no CH3ONO2 is seen in the catalyst output.
Figure 4 shows spectra for this case in a format similar
to Figure 3. In this case, the zeolite-based catalyst is present
downstream of the plasma, and there is no methyl nitrate
formed, as shown by the FTIR spectra.
When 46 ppm of CH3ONO2 is added to the input flow and
the gas is bypassed around the plasma/ catalyst, the gas
composition is as shown in column 6 of Table 3. The “sum
N species” value is increased by the nitrogen in the CH3-
ONO2 since all other gas flows were set to the same values
as before. When the gas is passed over the catalyst with plasma
power off (column 7), a small reduction in NO concentration
and a small production of formaldehyde are measured. The
CH3ONO2 is converted over the catalyst, at the same time
that 38 ppm of NO2 and 38 ppm of methanol are formed.
When the plasma power is on and the catalyst present
with added CH3ONO2, NO reduction is somewhat lower than
that without added CH3ONO2. NOx reduction (measured from
the input gas value) is reduced by a small amount. No CH3-
ONO2 is seen in the catalyst output, and the methanol
increases further.
FIGURE 3. FTIR spectra for plasma without catalyst (and without
added methyl nitrate). The upper trace is a portion of the spectrum
with the plasma off. The second trace is with the plasma power
on. The third trace is the result of subtracting power-on from power-
off spectra and then subtracting residual propene; this is a spectrum
of the products formed by the plasma. For comparison, the fourth
trace is a reference spectrum of methyl nitrate, and the fifth trace
is methanol.
with some of the HC appearing as formaldehyde, methanol,
and assorted other small molecular weight compounds. The
sum of the measured nitrogen species drops by about 20%.
Previous work (2) shows that this is not conversion to N2,
since a Cu-ZSM catalyst placed downstream of the plasma
brings NOx levels back to their input value. It should also be
noted that the same prior work showed that chemilumi-
nescent (CLA) measurements of NOx levels can be inaccurate
in this system due to interference from formaldehyde and
methanol.
Discussion
Figure 3 shows FTIR spectra for the cases of plasma power
off (top) and on (second trace), each without added methyl
nitrate. The power-off spectrum was subtracted from the
Several interesting conclusions can be drawn from the results
presented in Figure 3 and Tables 2 and 3. First, the IR spectra
9
VOL. 33, NO. 19, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 4 2 9

TABLE 3. Measured Gas Composition with Zeolite-Based Catalyst Comparing the Cases of Added CH ONO versus No Added
3
2
CH ONO , and Plasma Power off versus on
3
2
0 methyl nitrate
plasma
46 ppm methyl nitrate
plasma
component
formula
NO
NO2
N2O
HONO
“HC”
CH3ONO2
CH3OH
CH2O
off
on
conversion %
input gasa
off
on
conversion %
nitric oxide
nitrogen dioxide
nitrous oxide
acids
sum HC
m ethyl nitrate
m ethanol
222
17
2
87
49
10
61
234
1
2
4
2013
46
214
38
5
122
37
18
48
5
5
4
5
1927
b
b
14
1258
b
32
35
1910
b
38
16
1298
b
65
36
b
5
form aldehyde
144
136
sum N species
248
161
35
289
266
200
31
a
b
The “input” colum n is the m easured com position of the gas blend without passing through the oven/plasm a/catalyst system . Below detection
lim it.
°C (11). The competing loss process for CH3O radicals is
reaction with O2, which occurs with a rate constant of 5.3 ×
10-15 cm3 molecule-1 s-1 at 180 °C (11), i.e., a factor of 2000
less than k2. For the conditions of the present experiments,
[O2] ) 9% and [NO2] ≈ 100 ppm, the concentration ratio
[O2]/ [NO2] ) 900 and reaction with NO2 to give methyl nitrate
is likely to be an important loss process for CH3O radicals.
The germane question at this point is, what is the source of
CH3O radicals in the system? The answer to this question is
fairly straightforward. The simulated diesel exhaust gas
mixture contains a large amount (5%) of water vapor. Passage
of the H2O through the plasma will create H atoms and OH
radicals. OH radicals are well-known to react with hydro-
carbons to give alkyl radicals (R) which add O2 to give alkyl
peroxy radicals (RO2) which then react with NO to give alkoxy
radicals (RO). Alkoxy radicals undergo several different
reactions, including decomposition, which often generate
CH3 radicals which will then be converted to CH3O radicals
as described above. Almost all hydrocarbon fuels contain
substantial quantities of chemicals containing CH3 groups
and thus it should be expected that CH3ONO2 will be a
common product in a variety of plasma exhaust treatment
devices. Another question which arises now is, Is it likely
that other alkylnitrates (e.g., C2H5ONO2, C3H7ONO2, etc.) will
be formed in substantial yields? The answer here is probably
not, because the corresponding alkoxy radicals (C2H5O,
C3H7O, etc.) are expected to undergo rapid decomposition
via C-C bond scission rather than reaction with NO2 (13).
Second, the CH3ONO2 which is formed in the plasma is
not removed substantially as the gas passes through the
plasma. This observation is consistent with the relative
unreactivity of CH3ONO2 toward OH radicals. Under the
present experimental conditions (180 °C, one atmosphere)
OH radicals react 210 times faster with C3H6 and 30 times
faster with C3H8 than with CH3ONO2 [k5 ) 7.0 × 10-14 (11),
k6 ) 1.5 × 10-11 (11), and k7 ) 2.3 × 10-12 cm3 molecule-1
s-1 (12)]
FIGURE 4. FTIR spectra with plasma power off, power on, and
difference similar to Figure 3. In this case, the zeolite-based catalyst
is present downstream of the catalyst.
shown in Figure 3 show clearly that CH3ONO2 is formed in
a substantial yield on passing HC/ NOx/ O2/ N2 mixtures
through a nonthermal plasma. This has not been previously
reported. There are only two known gas-phase reactions by
which CH3ONO2 could be formed in the present system,
reactions 4b and 2:
OH + CH3ONO2 f products
OH + C3H6 f products
OH + C3H8 f products
(5)
(6)
(7)
CH3O2 + NO f CH3O + NO2
CH3O2 + NO + M f CH3ONO2 + M
CH3O + NO2 + M f CH3ONO2 + M
(4a)
(4b)
(2)
Third, the catalyst is very effective at removing CH3ONO2,
which appears to be converted into methanol and NO2.
Fourth, and most importantly, we have shown here that
CH3ONO2 is not an important intermediate in the NOx
removal reaction of the plasma-catalyst system. This leaves
open the question of what reactions in the plasma are critical
for NOx removal. With no evidence for the formation of C-N
The reaction of CH3O2 radicals with NO proceeds via two
channels with ,1% of the total reaction proceeding to give
CH3ONO2 (10). Reaction 4b cannot account for the CH3ONO2
observed in the present work. Reaction 2 occurs rapidly with
a rate constant of k2 ) 1.0 × 10-11 cm3 molecule-1 s-1 at 180
9
3 4 3 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 19, 1999

(5) Tonkyn, R.; Barlow, S. E.; Balmer, M. L.; Orlando, T. M.; Hoard,
J. W. Vehicle Exhaust Treatment Using Electrical Discharge
Methods; Society of Automotive Engineers paper number
971716, May 1997.
(6) Shelef, M. Selective Catalytic Reduction of NOx with N-free
Reductants. Chem. Rev. 1995, 95, 209-225.
(7) Penetrante, B. M.; Pitz, W. J.; Hsaio, M. C.; Merritt, B. T.; Voigtlin,
G. E. Effect of Hydrocarbons on Plasma Treatment of NOx;
Proceedings of the 1997 Diesel Engine Emissions Reduction
Workshop, University of CaliforniasSan Diego, July 28-31, 1997;
pp 123-127.
(8) Gierczak, C. A.; Andino, J. M.; Butler, J. W.; Heiser, G. A.; Jession,
J.; Korniski, T. J. FTIR: Fundamentals and Applications in the
Analysis of Dilute Vehicle Exhaust; SPIE-The International
Society for Optical Engineering, vol. 1433 Measurement of
Atmospheric Gases 1991, p 315ff.
bonds or other organic nitrates, we conclude that the N-N
bond formation occurs on the catalyst. Interestingly, prior
work (2) has shown that NOx reduction is poor on this catalyst
when NO2 is the input gas (i.e., in the absence of NO), and
thus it remains unclear which reactions are critical to NOx
removal.
Acknowledgments
The work presented here was performed at Ford Research
Labs as part of Ford’s participation in CRADA PNL-089 “New
Technologies for Exhaust Emission Reduction”, a cooperative
project of the USCAR Low Emission Technologies R&D
Partnership, Pacific Northwest National Lab, and the U.S.
Department of Energy.
(9) Blatt, A. H., Ed. Organic Synthesis; Coll., Wiley: New York, 1943;
Collect. Vol. 2, p 412-414.
(10) Wallington, T. J.; Dagaut, P. D., Kurylo, M. J. Chem. Rev. 1992,
Literature Cited
(1) Penetrante, Brusasco, Pitz, Vogtlin, Kung, Kung, Wan, and Voss
Plasma-Assisted Catalytic Reduction of NOx; Society of Automo-
tive Engineers paper number 982508, October 1998.
(2) Hoard, J. W.; Balmer, M. L. Analysis of Plasma-Catalysis for Diesel
NOx Remediation; Society of Automotive Engineers paper
number 982429, October 1998.
(3) Balmer, M. L.; Tonkyn, R.; Kim, A.; Yoon, S.; Jiminez, D.; Orlando,
T. M.; Barlow, S. E. Diesel NOx Reduction on Surfaces in Plasma;
Society of Automotive Engineers paper number 982511, October
1998.
92, 667.
(11) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.;
Molina, M. J. Jet Propulsion Laboratory Publication 97-4, 1997,
Pasadena, CA.
(12) Atkinson, R. Chem. Rev. 1986, 86, 69.
(13) Atkinson, R. Int. J. Chem. Kinet. 1997, 29, 99.
Received for review December 14, 1998. Revised manuscript
received June 2, 1999. Accepted June 16, 1999.
(4) Shimizu, S.; Oda, T. Catalyst Assisted Reaction by Using
Nonthermal Plasma on Nitric Oxide Removal. Denki Gakkai
Hoden Kenkyukai Shiryo; 1997; vol. ED-97, no. 209-225, p 39-
44 (Japanese language).
ES9813010
9
VOL. 33, NO. 19, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 4 3 1