Innovative surface wave plasma reactor technique for PFC abatement

Article abstract of DOI:10.1021/es9706514

Application of surface wave plasmas as an innovative technology for the destruction and removal of perfluorocompounds (PFC) emanating from semiconductor fabrication tools is demonstrated. The destruction of parts per thousand (ppt) concentrations of hexafluoroethane, C₂F₆, in oxygen and natural gas mixtures has been investigated as a function of microwave power in a low-pressure plasma reactor at 11.3 Torr. Effluent analysis included the determination of destruction and removal efficiencies (DRE) and product distributions by Fourier transform infrared spectroscopy and mass spectrometry. Destruction and removal efficiences of up to 99.6% for C₂F₆ were achieved using applied microwave powers from 500 to 2000 W, which corresponded to millisecond range residence times within the plasma. Product analysis indicated that hexafluoroethane conversion was limited to low molecular weight gases such as CO₂, CO, COF₂, H₂O, and HF. CF₄ was not produced as a plasma byproduct in any significant quantities. These investigations indicate that surface wave plasma destruction of perfluorocompounds at the point of use is a viable nonintrusive abatement technology for application to semiconductor manufacturing tools. Application of surface wave plasmas as an innovative technology for the destruction and removal of perfluoro-compounds (PFC) emanating from semiconductor fabrication tools is demonstrated. The destruction of parts per thousand (ppt) concentrations of hexafluoroethane, C₂F₆, in oxygen and natural gas mixtures has been investigated as a function of microwave power in a low-pressure plasma reactor at 11.3 Torr. Effluent analysis included the determination of destruction and removal efficiencies (DRE) and product distributions by Fourier transform infrared spectroscopy and mass spectrometry. Destruction and removal efficiencies of up to 99.6% for C₂F₆ were achieved using applied microwave powers from 500 to 2000 W, which corresponded to millisecond range residence times within the plasma. Product analysis indicated that hexafluoroethane conversion was limited to low molecular weight gases such as CO₂, CO, COF₂, H₂O, and HF. CF₄ was not produced as a plasma byproduct in any significant quantities. These investigations indicate that surface wave plasma destruction of perfluorocompounds at the point of use is a viable nonintrusive abatement technology for application to semiconductor manufacturing tools.

Full text of DOI:10.1021/es9706514

Environ. Sci. Technol. 1998, 32, 682-687
to global warming, but their projected growth could con-
tribute several percent to warming trends in the next century.
Perfluorocompounds such as C , CF , C , and SF are
Innovative Surface Wave Plasma
Reactor Technique for PFC
Abatement
2
F
6
4
3
F
8
6
removed very slowly from the atmosphere, and some of these
have estimated lifetimes greater than 10 000 years (2). In
addition, these perfluorocompounds are excellent absorbers
of infrared radiation and thus have very high global warming
potentials. As a result, effectively all such emissions ac-
cumulate in the atmosphere and will continue to influence
the climate for thousands of years. Currently, their con-
centrations and associated radiative forcings are low but they
may become significant in the future if unabated emissions
continue.
Perfluorocompounds are vital to the semiconductor
industry. The semiconductor industry incorporates numer-
ous low-pressure operations during the manufacturing of
semiconductor wafers such as chemical etch, which uses
perfluorocompounds. Integrated circuit or computer chip
manufacture is a $50 billion industry and is increasing at a
growth rate of 10-15% annually (3). It is estimated that
there are approximately 8000 of these tools in the United
States and double that number worldwide (3). Typically,
gaseous waste streams emanating from these process tools
that contain perfluorocompounds are diluted with vast
quantities of air or nitrogen and are either released into the
atmosphere or thermally incinerated. However, due to the
thermal stability of many of these perfluorocompounds such
C H R I S L . H A R T Z A N D J O H N W . B E V A N *
Department of Chemistry, Texas A&M University,
College Station, Texas 77843-3255
M A R C W . J A C K S O N A N D
B I L L A . W O F F O R D
Rf Environmental Systems, Inc., 1506 Loch Lake Drive,
Seabrook, Texas 77586
Application of surface wave plasmas as an innovative
technology for the destruction and removal of perfluoro-
compounds (PFC) emanating from semiconductor fab-
rication tools is demonstrated. The destruction of parts per
thousand (ppt) concentrations of hexafluoroethane, C2F6,
in oxygen and natural gas mixtures has been investigated
as a function of microwave power in a low-pressure
plasma reactor at 11.3 Torr. Effluent analysis included the
determination of destruction and removal efficiencies
2 6 4
as hexafluoroethane (C F ) and tetrafluoromethane (CF ),
(DRE) and product distributions by Fourier transform infrared
this incineration process is not completely effective, and thus,
some of these environmentally harmful compounds are
emitted into the atmosphere. Another disadvantage of the
thermal incineration method is that the waste streams from
numerous tool sets are diluted with nitrogen and combined
into a single waste stream prior to incineration. This leads
to numerous inefficiencies, because the incinerator must be
constantly operated at parameters that give the best results
for the bulk flow and not tuned to each individual stream.
The control of greenhouse gas emissions and the reduction
of potential global warming which would result are areas of
active global interest. Consequently, the semiconductor
industry has come under increased pressure to reduce
emissions of these compounds to the atmosphere. In
response to the U.S. Climate Change Action Plan, the industry
is actively evaluating various options such as alternative
chemistries, process optimization, recycle and recovery, and
abatement. While the first three options may be more
desirable from an environmental point of view, the hurdles
that have to be overcome for implementation are immense
due to the complex nature of the semiconductor manufac-
turing process. Currently, there are no obvious alternative
chemicals to replace perfluorocompounds. Abatement of
the emissions at the exhaust of the individual semiconductor
manufacturing tools appears to be closest to commercializa-
tion and closest to providing a cost-effective solution.
Recently, Bevan et al. have described a new novel plasma
device that operates in the radiofrequency range for the
destruction and removal of hard to incinerate volatile organic
compounds (4, 5). In this paper, we will describe a surface
wave plasma device that operates in the microwave region
and is extremely efficient for the destruction and removal of
perfluorocompounds, which are frequently used for semi-
conductor etching and cleaning operations. Specifically, we
have studied the low-pressure surface wave plasma reactions
of hexafluorethane with natural gas and molecular oxygen
in a tubular flow reactor as a function of absorbed microwave
frequency power. This study includes the determination of
spectroscopy and mass spectrometry. Destruction and
removal efficiences of up to 99.6% for C2F6 were achieved
using applied microwave powers from 500 to 2000 W,
which corresponded to millisecond range residence times
within the plasma. Product analysis indicated that
hexafluoroethane conversion was limited to low molecular
weight gases such as CO2, CO, COF2, H2O, and HF. CF4
was not produced as a plasma byproduct in any significant
quantities. These investigations indicate that surface
wave plasma destruction of perfluorocompounds at the point
of use is a viable nonintrusive abatement technology for
application to semiconductor manufacturing tools.
Introduction
It has recently become obvious that there is a discernible
human influence on global climate (1). Projections of future
global mean temperature change and sea level rise confirm
the potential for humanity to alter the earth’s climate to an
extent unprecedented in our brief history. The long time
scales governing both the accumulation of greenhouse gases
in the atmosphere and the response of the climate systems
to those accumulations mean that many important aspects
of climate change are effectively irreversible (2).
Increases in greenhouse gas concentrations since pre-
industrial times (i.e., 1750) have led to a warming of the
earth’s surface (See Table 1). The atmospheric concentration
of greenhouse gases such as CO
significantly, by about 30% and 145%, respectively (1892-
995). These trends can be attributed largely to human
2 4
and CH have grown
1
activities such as fossil fuel use. At the present time, some
long-lived greenhouse gases particularly PFCs contribute little
*
Corresponding author telephone: (409)845-2372; fax: (409)845-
4
719; e-mail: Bevan@chemvx.tamu.edu.
6
8 2
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 5, 1998
S0013-936X(97)00651-2 CCC: $15.00

1998 Am erican Chem ical Society
Published on Web 01/27/1998

TABLE 1. Concentrations and Global Warming Potentials (1, 2)
CO2
CH4
CF4
zero
72 pptv
2
C2F6
C3F8
pre-industrial concn
concn 1994
280 ppm v
358 ppm v
700 ppbv
1720 ppbv
0.62
zero
2 pptv
3.2
zero
zero
NA
rate of concn change (%/yr)
atm ospheric lifetim e (yr)
global warm ing potential (100 years)
global warm ing potential (500 years)
0.4
50-200
1
1
12
21
6.5
50 000
6 500
10 000
10 000
9 200
14 000
2 600
7 000
10 100
FIGURE 1. Schematic diagram of the surface wave plasma abatement system.
destruction and removal efficiencies for C
in-depth effluent analysis and mass balance.
2
F
6
as well as an
natural gas to the simulated waste stream. The addition of
a source of hydrogen allows a pathway for HF formation and
4
thus avoids the production of CF . For the duration of this
Experimental Section
study, the additive gases included additional molecular
oxygen (5 L/ min) and natural gas (3.6 L/ min) which was
mostly methane and ethane. These gases were also intro-
duced using MKS type 1179 variable rate mass flow control-
lers. The mixed gas system was then introduced into a 45
× 48 mm tubular quartz reactor. This quartz reaction vessel
passed perpendicularly through the surfaguide surface wave
launcher where a 2-kW short-duration microwave pulse
initiated a surface wave plasma within the quartz reaction
tube. After ignition, the microwave power supply returned
to its preset power level. The coarse microwave circuit tune
was accomplished by adjusting the sliding short circuit on
the surfaguide launcher. A three stub tuner attached to the
input of the surfaguide was then adjusted to maximize the
The abatement device used in this study is shown in Figure
. The surface wave high-frequency plasma discharge was
1
produced using a Sairem power supply, a microwave
generator at 2.45 GHz, a circulator, a three stub tuner, and
a surfaguide surface wave launcher with a sliding short circuit.
Semiconductor process gases (C
using MKS type 1179 mass flow controllers at a flow rate of
L/ min each into a Roots blower vacuum pump. After exiting
2 6 2
F and O ) were metered
1
the Roots blower, these process gases entered a mixing vessel
where additive gases were introduced to ensure the complete
destruction and removal of hexafluoroethane. Initially, only
additional oxygen was introduced into the waste stream.
However, the plasma reaction of C
tetrafluoromethane, CF , and large quantities of molecular
fluorine, F . CF has similar 100- and 500-year global warming
potentials and a higher lifetime global warming potential
than does C , so in order to hinder CF formation, we added
2 6 2
F and O produced
4
forward power level and minimize back reflections. A
2
4
circulator was placed between this three stub tuner and the
microwave generator to absorb any significant back reflec-
tion.
2
F
6
4
VOL. 32, NO. 5, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6 8 3

FIGURE 2. FTIR spectra illustrating the surface wave plasma destruction of PFC’s.
The inner quartz reaction tube resides inside a larger 74
79 mm tubular quartz containment vessel. A 15 scfm
process as well as to critically asses the viability of this
technology as a pollution control device for semiconductor
process tools.
×
stream of compressed air was flowed between this contain-
ment vessel and the reactor tube to prevent thermal
breakdown. The cooling air entered the system at room
temperature and exited the reactor system at 290 °C. The
entire plasma reaction system was contained in a Faraday
cage. This containment system served two purposes: (1)
prevent microwave radiation leakage, which is hazardous to
humans, and (2) prevent release of toxic or hazardous gases
in the event of catastrophic reaction tube failure. A Neslab
recirculating water chiller was used to cool the microwave
generator and circulator and the surface wave launcher.
The hexafluoroethane concentration was fixed throughout
Quantative gas-phase FTIR was performed using a Biorad
FTS-60 spectrometer utilizing a stainless steel heated cell
with a path length of 18 cm. A total of 100 co-added spectra
-1
were recorded from 500 to 5000 cm at a resolution of 1
-1
cm . The spectrometer was equipped with a potassium
bromide (KBr) beam splitter and an mecury-cadmium-
telluride (MCT) infrared detector. Again all transfer lines
were heated to minimize condensation and avoid adsorption
of analytes. Identification of HF, CO, H
2 2 4 2 6
O, CO , CF , C F ,
and COF was accomplished on the basis of rotational
2
constants, band origins, and by characterizing rotationally
resolved vibrational bands of standard reference molecules
chosen for calibration purposes. The absorbances of selected
rovibrational lines or unresolved rovibrational bands of
the study at 94 ppt. We chose 1 L/ min C
2 6
F flow because
typical recipe flows for C in semiconductor manufacture
2 6
F
vary from 200 to 1600 sccm. The molar ratios of the reactants
were fixed throughtout this study as
different concentrations of HF, CO, H
2 2 4 2 6
O, CO , CF , C F , and
COF gases in nitrogen were measured to generate calibration
2
1
C F + 3.6CH + 6O w
2
6
4
2
lines for each species. EPA Protocol 1 standard gases were
used when available to build a calibration matrix for the
FTIR system and to quantitate the observed spectra. The
concentrations of these gases in the unknown mixture were
determined by matching the absorbances of the selected
lines with the calibrated ones. EPA Protocols for the use of
Extractive Fourier Transform Infrared Spectrometry for the
Analyses of Gaseous Emissions from Stationary Sources formed
the basis for FTIR quantitative measurements (6).
This yielded concentations of C
O ) 566 ppt with a total reactor input flow of 10.6 L/ min.
2
F
6
) 94 ppt, CH ) 340 ppt,
4
2
The overall reactor system was leak checked using the on-
line Leybold Inficon mass spectrometer and kept under low
pressure utilizing an Edwards QDP80 dry vacuum pump with
a pumping speed of 65 cfm. Product gases exited the dry
pump and entered the fume hood where the acid gases were
water scrubbed.
Temperature measurements were made 1 m from the
microwave applicator. A Chromel-Alumel 304 SS sheath type
K 1/ 4-in. Omega quick disconnect thermocouple assembly
was used. A model HH-52 dual input switchable digital
thermometer manufactured by Omega Engineering was also
employed. Pressure measurements were made using a
Leybold Inficon IG3 vacuum gauge.
Product Analysis. In this work, initial reactant concen-
trations and final product distributions were determined for
various applied microwave powers. Quantative analysis of
the plasma reactants and byproducts were performed using
a variety of computer-controlled on-line and in situ analytical
techniques (see Figure 1). These included gas-phase Fourier
transform infrared spectrometry (FTIR) and mass spectrom-
etry (MS). Multiple analytical techniques were employed
for an in-depth understanding of the plasma destruction
Infrared spectra of the plasma destruction of C
given in Figure 2. One spectrum shows the components in
the gas stream before plasma application (C , CH , C
and O ), and the other spectrum shows the byproducts of
2 6
F are
2
F
6
4
2 6
H
,
2
the plasma reaction. Oxygen cannot be seen in the infrared
spectrum because it is not infrared active. Not only was the
destruction and removal efficiency for this experiment high
(
99.6%), but the process yields only simple, low molecular
weight byproducts such as CO, H O, CO , and other easily
scrubbed and neutralized compounds such as COF and HF.
2
2
2
Quantitative mass spectrometry was performed using a
differentially pumped in-line Leybold Inficon Transpector
200 AMU residual gas analyzer. This system consists
essentially of an ion source, quadrupole mass filter, Faraday
cup, and electron multiplier detectors. In-line mass spec-
trometry was used to sample the product stream at the outlet
6
8 4
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 5, 1998

FIGURE 3. These mass spectra demonstrate the high DRE and product distribution of the surface wave plasma abatement technology.
TABLE 2. Mass Spectral Cracking Patterns of Interest (8, 9)
peak 1
peak 2
peak 3
peak 4
peak 5
compd
mass
m/z
RA
m/z
RA
m/z
RA
m/z
RA
m/z
RA
C3F8
C2F6
CF4
COF2
OF2
CO2
F2
O2
C2H6
CO
188
138
88
66
54
44
38
32
30
28
20
18
16
2
69
69
69
47
35
44
19
32
28
28
20
18
16
2
100
100
100
100
100
100
100
100
100
100
100
100
100
100
31
119
50
66
54
16
38
16
27
12
19
17
15
1
8.8
41.3
11.8
55
62
9.4
55
9
33.6
4.7
169
31
19
28
16
28
24.6
18.3
6.7
14
44
8.2
50
50
31
31
19
12
9.3
119
12
25
9
10.1
4.9
4
10
6.7
1.5
3.9
3
50
28
30
16
1
3
24.1
1.7
26
29
23.5
1.2
29
14
21.2
0.8
HF
H2O
CH4
H2
21
85.6
2
14
17.1
13
8.6
12
2.6
of the plasma reactor and to avoid secondary gas-phase
reactions. A Welch DIRECTORR vacuum pump differentially
pumped the plasma reacted off-gas through a 1/ 8-in. circular
orifice where it held the input pressure to the quadrupole
mass spectrometer at 1 Torr. The Transpector 200 AMU
residual gas analyzer was equipped with an IPC28 pressure
converter which is an orifice system designed to allow the
mass spectrometer to sample the gas stream at 1 Torr. Four
liters per minute of product gas was pulled through the
residual gas analyzer by a Pfeiffer Balzers turbo molecular
pump, which was backed by a Leybold Heraus TRIVAC
vacuum pump. Sampling pressures in the mass spectrometer
previous experiments. The mass spectrometer was located
2 ft from the center of the plasma reactor and is only capable
of identifying stable products. The second mass spectrum
in Figure 3 shows the plasma byproducts attained by
operating the plasma abatement device at 2 kW. When the
microwave power is applied, the mass spectral features
belonging to C
2 6 2 2 6 4
F , O , C H , and CH decrease substantially,
and the spectral features due to the plasma byproducts arise.
+
+
+
+
+
These are CO
2
(44 CO
2
, 16 O , 28 CO , 12 C ), CO (28 CO ,
+
+
+
+
+
12 C ), COF
2
(47 COF , 66 COF
2
, 28 CO ), HF (20 HF , 19
O , 17 OH ). The possiblity of observing
and just F was pursued.
+
+
+
F ), and H
2
O (18 H
2
oxyfluorine compounds such as OF
2
2
-
5
-6
were typically 10 -10 Torr. As with the FTIR system,
standard gases were used to generate calibration curves for
the mass spectrometer (7). Mass spectrometric analysis did
not reveal the presence of any high molecular weight (C >
However, such compounds easily react with the moisture in
the effluent stream.
Results and Discussion
Initially, plasma studies were performed to destroy C F using
2 6
only molecular oxygen in a surface wave plasma device. The
results of this study are given in Table 3. The simplified
1
) organic species produced by the plasma reactor.
Figure 3 shows the two mass spectra that were taken
simultaneously with the aformentioned FTIR spectra. The
first mass spectrum shows the distinct cracking pattern (m/
unbalanced equation for the plasma destruction of C
additive oxygen is given below as eq 1:
2
F
6
using
+
+
+
+
+
2
z) for C
2
F
6
(69 CF
3
, 119 C
2
F
5
, 31 CF , 50 CF
2
+
), O
2
(32 O
,
+
+
+
+
1
6 O ), C
2
H
6
(28 C
2
H
4
, 27 C
, 15 CH
2
H
3
+
, 30 C
, 14 CH
2
H
6
+
, 26 C
, 13 CH , 12 C ).
2
+
H
2
, 29
+
+
+
C
2
H
5
), and CH
4
(16 CH
4
3
2
C F + O w COF + CF + CO + CO
(1)
2
6
2
2
4
2
The m/ z axis of these spectra were truncated at 70 for visual
+
clarity. The C
2
F
5
peak resulting from the ionization of C
2
F
6
at 119 m/ z is not shown. Mass spectral cracking patterns of
interest in this experiment are given in Table 2. The first
spectrum also shows residual water and carbon dioxide from
The main byproducts of this reaction in order of decreasing
concentration are COF > CF > CO :CO under the operating
conditions listed in Table 3. The destruction and removal
2
4
2
VOL. 32, NO. 5, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6 8 5

TABLE 3. Destruction and Removal of PFCs Using Oxygen
reactants
CF4
microwave power
W)
reactor pressure
(Torr)
C2F6
(L/min)
O2
(L/min)
C2F6
(ppt)
O2
(ppt)
production
(ppt)
emission
(L/min)
DRE
C2F6 (%)
(
1
1
1
1
1
1
1
950
950
950
950
950
950
950
2.47
2.87
3.21
3.52
4.26
4.72
5.33
1
1
1
1
1
1
1
1
1.5
2
3
4
500
400
333
250
200
167
143
500
600
667
750
800
833
857
144.8
58.5
17.8
6.4
4.3
3.2
0.36
0.19
0.07
0.03
0.03
0.02
0.02
96.5
97.5
98.0
98.1
95.7
84.8
70.4
5
6
2.3
TABLE 4. Destruction and Removal of PFCs Using Oxygen and Natural Gas
microwave
power
reactor
pressure
(Torr)
reactor residence
recovery
temp
(K)
time
(ms)
DRE
(ppt)
C2F6
(ppt)
CH4
(ppt) (ppt)
O2
CO2
(ppt)
CO
COF2
CF4
HF
H2O
(W)
(ppt) (ppt) (ppm) (ppt) (ppt) (ppt)
(%)
0
00
000
500
950
11.3
12.2
17.2
39.1
43.6
298.2
431.4
581.2
773.3
884.6
0
0
94
340
301
223
22
566
510
370
15 115
<1
0
3.6
25
0
30
71
160
184
0
<1
21
75
58
0
<1
<1
<1
0
27
90
0
1000 100
5
2.6
5.5
14
6.5 88
29.6 66
92.1
99.6
66 1026 103
153 1019 102
350
347
1
1
1
7.5
0.35
219
3.7 279
964
955
96
96
18.2
<1
87
efficiences for this process were a function of power and
oxygen concentration. DREs varied with oxygen concentra-
plasma abatement system yielding CO
2
, CO, COF
2
, H
2
O, and
HF. The overall experimentally determined balanced chemi-
cal equation for the 1950-W plasma destruction of hexafluo-
roethane is given below in eq 3:
tion and ranged from 70.4 to 98.1%. However, CF
was also significant at 2-145‰ (0.02-0.36 L/ min) depending
upon O :C ratio. At an applied microwave power of 1950
W, a C :O ratio of 1:3 gives the best destruction and removal
efficiency (98.1%) for C and the lowest CF production at
.4 ppt or 0.03 L/ min. Since CF is an extremely undesirable
4
production
2
2 6
F
2
F
6
2
1C F + 3.6CH + 6O w 1.3CO + 3.2CO + 1COF +
2
6
4
2
2
2
2
F
6
4
5
.2H O + 4HF (3)
2
6
4
byproduct from a global warming perspective, we added
natural gas to the reactant mixture to facilitate the destruction
and removal process. The addition of a suitable source of
This study was performed to determine the destruction and
removal as a function of power. The power requirements
for total destruction of PFCs are important because the cost
of the microwave power supply and generator is by far the
most expensive component of the abatement system. Thus,
the power requirements for complete abatement of PFCs
will determine the commercial viability of this process. In
addition, the major byproducts of the reaction were identified
and quantitated.
The actual hexafluoroethane destruction and removal
efficiencies (DREs) given in Table 4 were determined by
averaging independent quantative analyses obtained using
the two distinct (FTIR and MS) analytical methods. The FTIR-
and MS-determined DREs agreed with each other by less
than 0.5% in all cases. We have compared the hexafluoro-
ethane mass density (g/ mL) before and after entering the
plasma reactor to compute the correspondent DRE from its
standard definition (9):
4 2
hydrogen (CH or H ) forms a preferential pathway for the
formation of HF, a thermodynamically stable diatomic
species that limits the production of other fluorine containing
compounds.
The conversion of hexafluoroethane at a fixed input
concentration of 94 ppt was studied using different applied
microwave powers in a low-pressure (11.3-43.6 Torr) natural
gas and oxygen-containing surface wave plasma reactor. The
plasma column length was found to increase linearly with
applied power and decrease with increasing pressure. See
Table 4 for a summary of reactor conditions. Residence times
were computed as a function of plasma length and gas
velocity. The latter was determined using the following
expression (8):
QT
9
-
vf (m/ s) ) 5.31 × 10
(2)
(
)
Ap
Win - Wout
%
DRE )
× 100
(4)
(
)
Win
where Q is the total gas flow rate in L/ s, A is the cross sectional
2
area of the reactor in m , and T (K) and p (Pa) are the average
reactor temperature and pressure. In this study, the latter
were approximated to be the ones measured 1 m from the
wave launcher and ranged between 431 and 885 K depending
on specific reactor conditions. Residence times are also given
in Table 4. Residence times were extremely short ranging
from 2.6 to 18.2 ms.
This full-scale test was performed on this prototypical
abatement system in order to determine its longevity and
commercial viability. A simulated tungsten chamber clean
where W corresponds to the weight of C
2
F
6
in the mixture,
and the in and out subscripts denote the quantity before and
after being processed. The DREs for the microwave plasma
abatement of hexafluoroethane also given in Table 4 are
directly proportional to power and residence time. It is also
a strong function of the fractional composition of the waste
stream to be treated. The waste stream composition (process
gases + additive gases) used in this study was empirically
determined to be optimal for maximum destruction efficiency
as well as designed to produce the desired byproducts. In
source gas supply containing 1 L of hexafluoroethane (C
and 1 Lof oxygen (O ) was flowed into the off-gas abatement
sytem where it was mixed with 3.6 L of natural gas, mostly
CH and C , and 5 L of oxygen. The mixed feed and waste
gas stream were then passsed through the surface wave
2 6
F )
2
previous studies, excess oxygen was added to abate C
However, this process yielded CF as a byproduct. CF
known to have a higher lifetime global warming potential
than does C , the compound we were trying to abate. Thus,
2 6
F .
4
4
is
4
2 6
H
2 6
F
6
8 6
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 5, 1998

we decided to use additive gases such as CH
the formation of HF and to optimize the simultaneous
destruction and removal of CF . The extent of abatement is
4
or H
2
to promote
powers, and pressures. Moreover the process does not
require the addition of expensive rare gases nor preheating
of the reactant gas mixture. The byproducts of this abatement
device are all simple, low molecular weight compounds. The
hazardous byproducts can be easily water scrubbed. This
abatement device was also successful in the abatement of
4
also a complex function of the reactor pressure that is related
to inlet stream compostion and pumping speeds. Thus, the
efficiency of the reactor increased with increasing applied
microwave power and resulted in relatively large pollutant
removals (99.6%) when utilizing only 2 kW of incident
microwave power with a residence time of only 18.2 ms.
The product distributions from selected 500, 1000, 1500,
and 2000 W trial run analyses are summarized in Table 4.
These numbers are the average of the FTIR- and MS-
determined concentrations where applicable. The FTIR- and
MS-determined concentrations agreed with each other to
within 5% except for perfluorocompounds, which agreed to
within 1%. Not only is the destruction and removal efficiency
high, but the process also yields only simple, low molecular
weight byproducts. The striking simplicity of the species
produced by the plasma reaction (diatomic and triatomic
molecules) is in contrast to previously reported nonequi-
librium plasma studies. Hollahan (12) described the reactions
of saturated and unsaturated organic substances with
hydrogen and oxygen leading to a wide product distribution
of nonparent hydrocarbon species. This has been cor-
roborated by Barat and Bozzelli (13) in their low-pressure
reactions of chlorinated hydrocarbons under reducing and
oxidizing conditions. The main products of the surface wave
4 3 3 8
CF , NF , and C F with the aformentioned additive gases
(oxygen, natural gas, and hydrogen). In addition, as expected
in these oxygen-rich conditions, we did not observe any soot
formation within the reactor walls that would hinder
continuous operation. These aforementioned characteristics
make surface wave discharges an attractive alternative to
thermal incinerators for the destruction and removal of hard
to incinerate gases emanating from industrial processes
operating under low pressures such as semiconductor
fabrication tools.
Acknowledgments
This research is based in part upon work supported by The
Texas Advanced Research Program, NSF SBIR, and The Center
for Energy and Mineral Resources, Texas A&M University. In
addition, we thank SEMATECH for additional support.
Literature Cited
(
1) Climate Change 1995, The Science of Climate Change; Houghton,
J. T., et al., Eds.; Cambridge University Press: New York, 1996.
2) (a) Singh, H. B. Composition, Chemistry and Climate of the
Atmosphere; Van Nostrand: New York, 1995; p 218. (b) Harnisch,
J.; Borchers, R.; Maiss, M. Geophys. Res. Lett. 1996, 23, 1099.
(
plasma destruction are either relatively harmless (CO
O) or easily scrubbed and neutralized products (HF and
COF ). The mass recoveries for these trial runs were high at
6% and can be attributed to the use of multiple analytical
2
, CO,
H
2
2
(3) Sematech, private communication.
(
(
(
4) Arn o´ , J.; Bevan, J.; Moisan, M. Environ. Sci. Technol. 1995, 29,
961.
5) Arn o´ , J.; Bevan, J.; Moisan, M. Environ. Sci. Technol. 1996, 30,
427.
9
1
methods and techniques.
The product distributions resulting from the surface wave
plasma reactions that we have investigated are likely to stem
from both thermal and nonthermal processes. During the
nonthermal stage, the species travel throughout the plasma
zone and are subjected to a number of electron collisions in
a time range of 1-10 ms. Unlike plasma processes occurring
in the few Pa pressure range, the neutral gas temperatures
within the plasma at the reactor pressures used in our
investigation are estimated between 1000 and 2000 K
depending on the power applied, reactor pressure, and gas
composition (14). Thus, in the second stage, the stable and
intermediate species generated in the first stage experience
a fast cooling (ca. 100 K/ ms) as they exit the plasma volume.
The resulting short residence times are likely to limit the
number of slow multistep mechanisms that are essential in
the formation of more complex nonparent fluorohydrocarbon
species.
2
6) Zazzera, L. A.; Reagan, W. K. J. Electrochem. Soc. 1997, 144,
3597.
(7) Quantitative Interpretation of Mass Spectra. 2nd ed.; Leybold
Transpectra Gas Analysis System Manual, Chapters 4.3 and 4.4;
Leybold Inficon: Syracuse, New York, 1993.
(
8) Stanhagen, E.; Abrahamsson, S.; McLafferty, F. W. Atlas of Mass
Spectral Data; John Wiley & Sons, Inc: New York, 1969; Vols 1
and 2.
(9) Eight Peak Index of Mass Spectra, 2nd ed.; Mass Spectrometry
Data Centre: U.K. 1974; Vol. 1, Table 1.
(
(
10) Granier, A. In Microwave Discharges; NATO ASI Series: B 302;
Moisan, M.; and Ferreira, C. M., Eds.; Plenum Press: New York,
1
993; p 491.
11) Brunner, C. R. Hazardous Air Emissions from Incinerators;
Chapman and Hall: New York, 1985.
(12) Hollahan, J. R. In Techniques and Applications of Plasma
Chemistry; Hollahan, J. R., Bell, A. T., Eds.; John Wiley & Sons:
New York, 1974.
(
(
13) Barat, R. B.; Bozzelli, J. W. Environ. Sci. Technol. 1989, 23, 666.
14) Brown, L. C.; Bell, A. T. Ind. Eng. Chem. Fundam. 1974, 13, 203.
In conclusion, this study demonstrates the feasibility and
advantages of utilizing low-pressure surface wave plasmas
to destroy and remove thermally stable gases such as
hexafluoroethane emanting from semiconductor fabrication
Received for review July 28, 1997. Revised manuscript re-
ceived November 13, 1997. Accepted November 17, 1997.
tools. The reactor has been shown to efficiently destroy C
2 6
F
under conditions of low residence times, applied microwave
ES9706514
VOL. 32, NO. 5, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6 8 7