Chemosphere 76 (2009) 1273–1277
Contents lists available at ScienceDirect
Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere
Technical Note
Assessment of removal efficiency of perfluorocompounds (PFCs)
in a semiconductor fabrication plant by gas chromatography
Chang-Feng Ou Yang a, Seak-Hong Kam a, Chia-Hung Liu b, Jiren Tzou b, Jia-Lin Wang a,*
a
b
Department of Chemistry, National Central University, Chung-Li 320, Taiwan
Intraco Taiwan Corporation, 5F-2, No. 22, St. Tai-Yuan, Chu-Bei 302, Taiwan
a r t i c l e
i n f o
Article history:
Received 11 March 2009
Received in revised form 11 June 2009
Accepted 12 June 2009
Available online 14 July 2009
Keywords:
Local scrubber
Chemical vapor deposition (CVD)
Semiconductor
Global warming potential (GWP)
Destruction or removal efficiency (DRE)
a b s t r a c t
This study investigated a gas chromatographic (GC) method to assess the destruction or removal efficiency (DRE) of local scrubbers on five perfluorocompounds (PFCs), i.e., SF6, NF3, CF4, C2F6, and C3F8, which
are very potent greenhouse gases used in a semiconductor fabrication plant. Air samples taken at inlets
and outlets of local scrubbers were analyzed by a self-constructed multi-column GC system equipped
with thermal conductivity detection. Three packed columns were integrated into the heart-cut GC system
to allow simultaneous analysis of the five target PFCs. The Porapak Q pre-column performs rough separation and cuts eluent groups to two analytical columns for optimal separation. The Molecular Sieve – 5A
column separated NF3, CF4, and C3F8 and the second Porapak Q separated SF6 and C2F6. Linearity was
greater than 0.995 (R2) for the five PFCs, and the reproducibility was about 4% (relative standard deviation) for NF3, and better than 0.5% for the other four PFCs.
DRE for the combustion (CB) and electric–thermal types of local scrubbers was evaluated by taking into
account the in-line dilution from air and fuel gases. Both flow and tracer methods were employed to
deduce the dilution factors (DFs). For the tracer method, helium was employed as the tracer and injected
upstream of the scrubbers and thus mixed with the exhaust gas. With this method, the DFs were determined to be in the range from 4.8 to 5.9 for the CB unit, significantly higher than the value of 3.3 based on
the flow method. The DREs for the CB unit for C3F8 were greater than 90% and between 40% and 50% for
CF4.
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
Technological advances in semiconductor devices that increase
their throughput (number of devices per chip) and the circuit
speed are in large part attributable to contamination reduction
and prevention. Contamination can be introduced in various steps
of a semiconductor manufacturing process, resulting in reduction
in yields and electrical performance. In situ cleaning of chemical
vapor deposition (CVD) chambers is a process designed to reduce
the formation of silicate particles that can cause low yields by
depositing on the silicon wafers. Semiconductor and liquid crystal
display panel manufacturers also use CVD to create intricate circuitry patterns on silicon wafers or glass subtracts. Both in situ
cleaning and etching processes use perfluorocompounds (PFCs)
to produce fluorine-containing radicals inside the CVD reactor.
PFCs are usually very stable, due to the strong carbon fluoride
bonding. Other fully fluorinated compounds such as SF6 and NF3
also have similar properties. Their critical role in the electronics
industry has led to rapid growth in consumption since their intro* Corresponding author. Tel.: +886 3 4227151x65906; fax: +886 3 4277972.
E-mail address: [email protected] (J.-L. Wang).
0045-6535/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chemosphere.2009.06.039
duction in the 1980s (US EPA, 2000). Other industries such as aluminum production also contribute to the release of CF4 and C2F6
due to the anodic effect (Weston, 1996; Kokorin and Rand, 2000;
US EPA, 2000; Tsai et al., 2002). These chemicals are generally
low in toxicity and non-flammable, and their minute presence in
ambient air is not harmful from health and safety point of view
(Tsai et al., 2002). Nevertheless, the extremely high stability of
PFCs prevents their photochemical destruction in the troposphere
and, hence, results in their accumulation in the atmosphere (Ehhalt
and Prather, 2001; Khalil et al., 2003). The consequence is the
enhancement of the greenhouse effect due to their extremely high
global warming potentials (GWPs) (Ehhalt and Prather, 2001). As a
result, controlling their use in industry has become increasingly
important, as concern over global warming issues has increased.
Due to the limited efficiency of PFC use in etching or cleaning processes (approximately 10–80%), mitigation after the point-of-use is
necessary to reduce emissions into the atmosphere. For this purpose, the local scrubbers based on electric–thermal, plasma, fuelinjected combustion, or catalytic decomposition, are installed after
the CVD (Chang and Chang, 2006). Assessment of the destruction
or removal efficiency (DRE) of local scrubbers is essential for the
determination of emission factors for PFCs (Kokorin and Rand,
1274
C.-F. Ou Yang et al. / Chemosphere 76 (2009) 1273–1277
2000). The best-known methods for assessing PFC emissions are
Quadrupole Mass Spectrometry (QMS) and Fourier Transform
Infrared Spectroscopy (FTIR) (Wofford et al., 1999; Yu and Chang,
2001; Hong and Uhm, 2003). The major advantage associated with
these two techniques is real-time and simultaneous measurements
of multiple PFCs, which can provide instantaneous emissions of a
CVD with adequate sensitivity. Although gas chromatography
(GC) is extremely suitable for separating complex gas mixtures,
its slow analysis speed makes it inadequate for instantaneous
determination. This is particularly true when the concentrations
of PFCs and reaction products in the CVD vary with time such as
during a cleaning process (Zazzera et al., 1997). Nevertheless, the
need for instantaneous determination becomes less stringent for
the assessment of the DREs of local scrubbers, where the ratios
of outlet to inlet concentrations are to be determined, rather than
the instantaneous concentration versus time. In this case, GC techniques may be an effective substitute for FTIR or QMS. The low
cost, ruggedness, and high separation power of GC compared to
FTIR or QMS make it extremely appealing in DRE assessment when
speed is less of a concern.
Successful GC analysis of PFCs has been reported in the literature (Greene and Wachi, 1963; Bright and Matula, 1968; O’Mahony
et al., 1993; Culbertson et al., 2000) with either packed or capillary
columns. Detection methods for PFCs are commonly performed by
mass spectrometry, helium ionization, or thermal conductivity,
with the first two methods having much greater sensitivity than
the third (Andrawes et al., 1980).
In this study, a series of packed columns were tested and configured into a multi-column chromatographic system for assessing
the DREs of local scrubbers targeting five PFCs used in a semiconductor fabrication plant; they are NF3, SF6, CF4, C2F6, and C3F8. Gas
samples were taken from the inlets and outlets of two types of
local scrubbers, followed by GC analysis. Because air and fuel gases
are injected into local scrubbers for combustion or reaction
purpose, the resulting dilution effect on DREs of PFCs needs to be
accounted for when assessing DREs. A method of chemical tracer
for the dilution factor (DF) determination involving the use of
helium gas was also proposed and demonstrated.
2. Experimental
Because of the gaseous nature of the five target PFCs, GC was
used. Four different packed columns, i.e., Molecular Sieve 5A
(MS-5A, mesh size: 60/80, length = 2 m), Molecular Sieve 13X
(MS-13X, mesh size: 100/120, length = 1.8 m), Silica Gel (mesh
size: 100/120, length = 2 m) and Porapack Q (mesh size: 80/100,
length = 2 m), were tested for separation. Column packing was
done by pre-cutting a desired length of stainless steel tubing
(2 mm ID, 3.2 mm OD) with an electrically polished inner surface.
The empty stainless steel tube was plugged at one end by glass
wool and was connected to a pump to create a partial vacuum,
while packing material was poured into the other end. Filling
was performed while knocking on the tubing to facilitate tight
packing. A glass wool plug was applied to the other end to hold
the packing material in place. Conditioning was performed for a
few hours at an oven temperature of 150 °C under the flow of high
purity nitrogen.
The system design is similar to previous research (Wang et al.,
2001). Two sample loops were employed for gas injection. The
5.6 mL sample loop was used for real samples taken from the exhaust pipelines, whereas the micro-sample loop (a few lL) was
used for injecting pure or standard mixtures of PFCs for testing
separation. The sample loop was evacuated by a mechanic pump
prior to sample loading. Subsequently, the sample loop was flushed
and filled to the desired pressure (setpoint = 93.3 kPa) controlled
He
tracer
PFCs
CVD
Chamber
Auxiliary
Gases
Local
Scrubber
Pump
Inlet
Sampling
Central
Scrubber
Outlet
Sampling
Fig. 1. Illustration of a typical pipeline system and configuration of the inlet/outlet
sampling ports.
by a capacitance diaphragm gauge (622A13TAE, MKS, USA). The
sample loop was thermally controlled at 40 °C. A 2-way, 6-port
switching valve connected the sample loop and the transfer line
for sample injection into the packed columns in a GC equipped
with thermal conductivity detection (TCD, Agilent 6890, USA) kept
at 200 °C. The temperature program was kept the same for the four
tested columns: 4.0 min at 40 °C, an increase of 35 °C min1 to
150 °C, and then 4.0 min at 150 °C. The flow rate of the carrier
gas (high purity helium) was set at 30 mL min1 for all tested
columns.
Sampling was performed by connecting polytetrafluoroethylene
tubes (6.35 mm OD) to the inlet and outlet sampling ports of local
scrubbers (Fig. 1). The exhaust gas was drawn to the 2 L Tedlar
bags by a diaphragm pump (Model 2380, On-Line Tech., USA).
For standard preparation, pure PFC gases were diluted to 2000–
3000 ppmv by zero air with a vacuum line and the standard gas
mixture was stored in Tedlar bags.
Two types of local scrubbers were tested for their DREs in this
study, hereafter termed combustion (CB) unit and electric–thermal
(ET) unit, respectively. To validate the GC results, on-line FTIR
(Model 2010, On-Line Technologies, USA) was also employed to acquire instantaneous measurements.
3. Results and discussion
Because concentrations of the few gaseous PFCs and their byproducts in the exhaust pipelines are generally high (>ppm level),
using packed columns to allow large sample throughput compensates for the low sensitivity of thermal conductivity detection.
Although using capillary columns can provide better resolution,
their small throughput limits the detection by TCD; and, thus,
other detection methods need to be sought, which increases the
cost and fragility of the system. The combination of packed columns and TCD makes the analytical system extremely heavy-duty
and thus suitable for in-plant on-line or off-line applications.
Fig. 2 displays the superimposed chromatograms from the four
tested columns by injecting individual pure PFCs into the microsample loop. No single column can provide baseline separation
for the five PFCs. Nevertheless, the Porapack Q column revealed
overall the best separation of the target PFCs, which agrees with
the findings by Bright and Matula (1968), except that CF4 is coeluted with NF3. CF4 is a critical compound to be measured in
the exhaust because it has large GWP and can be either a reagent
or a by-product in the CVD (Koike et al., 1997). The MS-5A and
Porapack Q columns revealed complementary results (see Table 1
for the summary of the four columns). As a result, the MS-5A
and the Porapack Q combined appeared to be the most useful to
separate the five PFCs.
A dual-phase three-column system was therefore built to
simultaneous analyze the five PFCs in the CVD exhaust simultaneously. Although simultaneously injecting all five PFCs in one
cleaning recipe is rare, full separation is still desired to correspond
with the various needs of a plant, which calls for different PFCs for
1275
C.-F. Ou Yang et al. / Chemosphere 76 (2009) 1273–1277
(b)
(a)
Retention time (min)
Retention time (min)
(d)
(c)
Retention time (min)
Retention time (min)
Fig. 2. Superimposed chromatograms from overlaying individual chromatograms of five PFCs for (a) Porapak Q; (b) silica gel; (c) MS-13X; and (d) MS-5A.
Table 1
Separation conditions on five PFCs by four types of columns.
PFC
Silica gel
Porapak Q
MS-13X
MS-5A
CF4
C2F6
C3F8
NF3
SF6
X
X
O
X
X
X
O
O
X
O
X
X
X
X
X
O
X
O
O
X
‘‘X” denotes not separated; ‘‘O” denotes separated.
different CVD cleaning processes. Reaction by-products also occur
by plasma induced radical reactions. For instance, CF4 or COF2 may
be produced when C2F6 or C3F8 is used as the CVD cleaning agent
(Zazzera et al., 1997; Wofford et al., 1999; Yu and Chang, 2001).
Thus, the versatility of an analytical method mostly lies in its ability to separate most of the PFCs in service in a plant. Fig. 3 shows
the configuration of the two-phase three-column system. A short
Porapack Q column (length = 1.2 m, 2 mm ID) was used as the
pre-column to provide rough separation. The early eluting peak
group (i.e., air, NF3, and CF4) was further cut to the MS-5A
(length = 2.4 m, 2 mm ID), and the remaining three compounds
(i.e., C2F6, SF6, and C3F8) were cut to a second Porapack Q column
(length = 2.5 m, 2 mm ID) for improved separation. The cutting
was executed by a 2-way 4-port switch valve (A2C4UWT, Valco,
USA). Although the separation was largely accomplished, the peak
of C3F8 showed severe tailing (result not shown), which could result in large analytical uncertainty. The co-elution with water vapor was found to be the major cause of such tailing. Water vapor
is abundant in the exhaust pipes, and because such interference
is difficult to avoid, water vapor is often the major source of interference in FTIR measurements. This interference is less prevalent in
the MS column compared to the Porapack Q column, possibly because MS is a good absorber of water vapor itself. Consequently,
the last eluting compound (i.e., C3F8) from the pre-column was
not eluted from the Porapack Q column; rather, it was cut to the
MS-5A column again. The result is that SF6 and C2F6 were eluted
from the Porapack Q column, and CF4, NF3, and C3F8 were eluted
from the MS-5A. The major air matrix component, i.e., N2 and O2,
were also resolved by the MS-5A column (see Fig. 4 for the
completed chromatograms).
Fig. 3. Schematic diagram of the three-column system.
1276
C.-F. Ou Yang et al. / Chemosphere 76 (2009) 1273–1277
(a)
(b)
Fig. 4. Chromatograms of the three-column system separating five target PFCs.
The complete system was then tested for linearity and reproducibility before being applied to the assessment of the DRE of
local scrubbers. In the assessment of the DRE of a local scrubber,
the ratio of the outlet response to the inlet response is proportional
to the DRE, which is defined by
DRE ¼ 1 Outlet response Dilution factor
Inlet response
ð1Þ
The linearity of the method is critically important in order to accurate estimate the outlet/inlet response ratios and the DRE, after taking into account the DF, discussed further below. The test of
linearity is performed by stepwise increasing the pressure of the
standard mixture in the sample loop containing the five PFCs balanced by air. Based on the ideal gas law, pressure is proportional
to the number of moles at constant volume (i.e., sample
loop = 5.6 mL) and temperature (i.e., sample loop thermally stated
at 40 °C). Thus, a linear relationship for the response with respect
to the pressure should be expected. In this research, eight different
pressures from 13.3 to 106.7 kPa were produced to generate
calibration curves for the five PFCs. Concentrations can be deduced
by using the pressure setpoint for routine measurements (93.3 kPa)
as the reference point. Normalized equivalent concentrations (%)
from 0.19% to 1.51% are thus for the pressure from 13.3 to
106.7 kPa. As expected, the relationship was highly linear, with an
R2 greater than 0.998. See Supplemental materials for detailed
information.
Analytical precision was tested by repeatedly injecting 10 aliquots of the standard mixture, and the relative standard deviation
(RSD) was better than 0.5% for CF4, SF6, C2F6, and C3F8. The RSD was
poorer for NF3 (P4%) because of the broad peak shape.
For the CB or ET types of local scrubbers, fuel and clean air are
injected for the decomposition of PFCs. Thus, dilution of PFCs arising from the influx of air and fuel (termed auxiliary gases) also
takes place at the same time when chemical decomposition occurs,
which contributes an additional decrease in PFC concentrations. As
a result, a DF needs to be taken into account and offset in the DRE
calculation (see Eq. (1)). For FTIR measurements, the DRE is usually
determined by shutting down the heating or combustion operation
of the local scrubber while maintaining the fuel and air flows
(Radoiu, 2004). No destruction occurs under this condition, and
the observed decrease in measurement response is solely due to
dilution. The drawback of this method is that the DF is not
determined in situ and, hence, the dilution effect determined in
this manner may not represent the true operational condition. Calculating the DF by flow rates of all gas streams in and out of the
scrubber is an alternative (i.e., flow method); however, flow meters
for all steams must be consistently calibrated to avoid bias. Giving
the above limitations, a simple solution is to use a chemical tracer
that cannot be decomposed by the local scrubber and to look for its
extent of decrease in response. However, few chemical compounds
can sustain the high operating temperatures in local scrubbers and
be detected by FTIR. Lee et al. (2007) have proposed to use He as an
internal standard to correct the gas flow rate for plasma scrubbers
during the manufacturing process. Here we also used He as the tracer being subjected to the same conditions as the PFCs in the pipelines for the CB and ET type scrubbers. These mono-atomic
molecules, like He or Ar, are IR-inactive and hence cannot be
exploited by the FTIR technique. With the GC technique, however,
these inert molecules can be detected by TCD and separated from
O2, N2, and other PFCs, provided that proper chromatographic conditions are met. Either He or Ar can be used as the tracer. Because
Ar was co-eluted with the air peaks, consequently He was chosen
to be the tracer for the DF determination. Helium was injected
approximately 1 m upstream of the inlet sampling port at a flow
rate of 100 and 300 mL min1 to facilitate mixing with the exhaust
before sampling from the inlet and outlet ports (Fig. 1).
Two types, three units of local scrubbers were tested for DRE of
C3F8 while a plant is at the evaluation stage of replacing CF4 or C2F6
by C3F8 for better fluoride utilization efficiency and lower
emissions (Zazzera et al., 1997). Both flow and tracer methods
were conducted to account for the DF. Two different scenarios of
high injection flow (900 mL min1) versus low injection flow
(650 mL min1) of C3F8 were carried out for all tested scrubbers.
As an example for C3F8, a high DRE (>96%) was found for the CB
unit and considerably lower DREs were found for the two ET units
(<45%), after taking into account the DF (3.3) determined by the
flow method (calculation not shown). Although only C3F8 was used
for testing, CF4 was also produced due to radical reactions in the
plasma. An unknown peak was found at the retention time of
about 7 min for the CB type, which is suspected to be COF2, possibly also resulting from the cleaning process, as supported by the
FTIR measurements (Wofford et al., 1999). Similar results were also
obtained by the on-line FTIR measurements, revealing relatively
higher removal efficiency for the CB unit and lower value for the
ET unit (see Supplemental materials).
Because the CB type was found to have much greater DREs than
the ET type for C3F8, whether the DF estimated by flow calculation
was accurate remains doubtful. As a result, the He tracer method
was attempted to validate the DF for the same CB unit. In Fig. 5, helium is visible in the chromatograms and is much lower for the
Fig. 5. DRE test for C3F8 in high flow cleaning of a CB unit with He as a tracer. The
blue trace corresponds to the inlet measurement, and the red trace corresponds to
the outlet measurement. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
C.-F. Ou Yang et al. / Chemosphere 76 (2009) 1273–1277
Table 2
DREs of C3F8 for the CB type unit using He as the tracer to account for DF.
Inlet (peak area)
Outlet (peak area)
DF
1374
70,541
6546
282
735
650
4.88
CB type; high flow cleaning mode
He (300 mL min1)
3600
49,763
C3F8
6422
CF4
751
662
820
4.79
CB type; low flow cleaning mode
4658
He (300 mL min1)
48,551
C3F8
6743
CF4
788
647
600
5.91
He (100 mL min
C3F8
CF4
1
)
1277
Acknowledgements
DRE (%)
95
52
This research was supported by the National Science Council of
Taiwan under the contracts of NSC97-2752-M-008-006-PAE and
NSC97-2111-M-008-002-AGC.
Appendix A. Supplementary material
94
39
92
47
outlet measurement than the inlet, confirming that considerable
dilution did occur with the injection of an additional air stream
into the scrubbers. Table 2 shows the complete DRE results with
the DF values determined by the He tracer method. The DF values
were very consistent for the two high flow rate cleaning trials (4.9
and 4.8). The DF value was slightly higher for the low flow rate
cleaning trial (5.9). These DF values were generally higher than
the DF value based on the flow method (DF = 3.3). In this case,
the discrepancy in DF does not induce considerable error for compounds with DREs close to unity (near complete destruction). For
instance, the DRE of C3F8 for the CB unit is only re-adjusted slightly
from >96% for the flow method to 92–95% for the tracer method.
The discrepancy could become significant for compounds with
moderate DREs. For instance, CF4 has DRE of 47% for the low flow
cleaning trial shown in Table 2. A large discrepancy in DF, say, from
5.9 to 3.3, could result in an underestimate of the DRE by 44%, lowering the DREs to only 26% for CF4. The tracer technique, which
provides more realistic determination of DF and possibly better
accuracy in DRE estimates, further demonstrates the advantage
of the GC technique in PFC emission assessment.
4. Conclusions
A dual-phase three-column GC-TCD method was developed for
measuring five target PFCs (CF4, C2F6, C3F8, SF6, and NF3) in the exhaust of CVD tools in a semiconductor fabrication plant. Heart-cuttings of the eluents from the Porapak Q pre-column to a MS-5A and
a second Porapak Q were configured to separate the five PFCs. Linearity denoted by the R2 was greater than 0.998 for the five PFCs,
and the reproducibility denoted by the RSD was about 4% for
NF3, and better than 0.5% for the other four PFCs.
DREs were determined for CB and ET types of local scrubbers.
DF values for the local scrubbers were taken into account based
on both flow and tracer methods. Using He as the tracer, the DFs
were determined to be significantly higher than the values from
the flow method (5–6 versus 3). Based on the GC and the tracer
methods, the DREs for the CB units were greater than 90% for
C3F8 and between 40% and 50% for CF4.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.chemosphere.2009.06.039.
References
Andrawes, F.F., Gibson, E.K., Bafus, D.A., 1980. Analysis of low molecular weight
perfluoroalkanes by gas chromatography with helium ionization detection.
Anal. Chem. 52, 1377–1379.
Bright, R.N., Matula, R.A., 1968. Gas chromatographic separation of low molecular
weight fluorocarbons. J. Chromatogr. 35, 217–222.
Chang, M.B., Chang, J.-S., 2006. Abatement of PFCs from semiconductor
manufacturing processes by nonthermal plasma technologies: a critical
review. Ind. Eng. Chem. Res. 45, 4101–4109.
Culbertson, J.A., Prins, J.M., Grimsrud, E.P., 2000. Improvements in the detection and
analysis of CF3-containing compounds in the background atmosphere by gas
chromatography-high-resolution mass spectrometry. J. Chromatogr. A 903,
261–265.
Ehhalt, D., Prather, M., 2001. Atmospheric chemistry and greenhouse gases. In: Joos,
F., McFarland, M. (Eds.), Climate Change 2001: The Scientific Basis. IPCC,
Cambridge, UK, pp. 241–287.
Greene, S.A., Wachi, F.M., 1963. Separation of some low molecular weight
fluorocarbons by gas chromatography. Anal. Chem. 35, 928–929.
Hong, Y.C., Uhm, H.S., 2003. Abatement of CF4 by atmospheric-pressure microwave
plasma torch. Phys. Plasmas 10, 3410–3414.
Khalil, M.A.K., Rasmussen, R.A., Culbertson, J.A., Prins, J.M., Grimsrud, E.P., Shearer,
M.J., 2003. Atmospheric perfluorocarbons. Environ. Sci. Technol. 37, 4358–4361.
Koike, K., Fukuda, T., Fujikawa, S., Saeda, M., 1997. Study of CF4, C2 F6, SF6 and NF3
decomposition characteristics and etching performance in plasma state. Jpn. J.
Appl. Phys. 36, 5724–5728.
Kokorin, A., Rand, S., 2000. PFC, HFC, SF6 emissions from primary semiconductor
manufacturing. In: Rosland, A. (Ed.), Good Practice Guidance and Uncertainty
Management in National Greenhouse Gas. IPCC, Montreal, Canada, pp. 69–78.
Lee, J.Y., Lee, J.B., Moon, D.M., Souk, J.H., Lee, S.Y., Kim, J.S., 2007. Evaluation method
on destruction and removal efficiency of perfluorocompounds from
semiconductor and display manufacturing. Bull. Kor. Chem. Soc. 28, 1383–1388.
O’Mahony, T.K.P., Cox, A.P., Roberts, D.J., 1993. Gas-chromatographic separation of
perfluorocarbons. J. Chromatogr. 637, 1–11.
Radoiu, M.T., 2004. Studies on atmospheric plasma abatement of PFCs. Radiat. Phys.
Chem. 69, 113–120.
Tsai, W.T., Chen, H.P., Hsien, W.Y., 2002. A review of uses, environmental hazards
and recovery/recycle technologies of perfluorocarbons (PFCs) emissions from
the semiconductor manufacturing processes. J. Loss Prevent. Proc. Ind. 15, 65–
75.
US EPA, 2000. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990–1998.
US EPA, Washington, DC, USA.
Wang, J.-L., Kuo, S.-R., Ma, S.-S., Chen, T.-Y., 2001. Construction of a low-cost
automated chromatographic system for the measurement of ambient methane.
Anal. Chim. Acta 448, 187–193.
Weston, R.E., 1996. Possible greenhouse effects of tetrafluoromethane and carbon
dioxide emitted from aluminum production. Atmos. Environ. 30, 2901–2910.
Wofford, B.A., Jackson, M.W., Hartz, C., Bevan, J.W., 1999. Surface wave plasma
abatement of CHF3 and CF4 containing semiconductor process emissions.
Environ. Sci. Technol. 33, 1892–1897.
Yu, S.J., Chang, M.B., 2001. Oxidative conversion of PFC via plasma processing with
dielectric barrier discharges. Plasma Chem. Plasma Process 21, 311–327.
Zazzera, L., Reagen, W., Cheng, A., 1997. Infrared study of process emissions during
C3 F8/O2 plasma cleaning of plasma enhanced chemical vapor deposition
chambers. J. Electrochem. Soc. 144, 3597–3601.