Journal of Hazardous Materials 294 (2015) 41–46

Contents lists available at ScienceDirect

Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat

Abatement of fluorinated compounds using a 2.45 GHz microwave

plasma torch with a reverse vortex plasma reactor
J.H. Kim a , C.H. Cho a , D.H. Shin a , Y.C. Hong a,∗ , Y.W. Shin a,b
a
Plasma Technology Research Center, National Fusion Research Institute, 814-2 Oxikdo-dong, Gunsan-city, Jeollabuk-do, South Korea
b
School of Advanced Green Energy and Environments, Handong Global University, Heunghae-eup, Buk-gu, Pohang-city, Gyeongbuk, South Korea

h i g h l i g h t s

• We developed a microwave plasma torch with reverse vortex reactor (RVR).

• We calculated a volume fraction and temperature distribution of discharge gas and waste.
• The performance of reverse vortex reactor increased from 29% to 43% than conventional vortex reactor.

a r t i c l e i n f o a b s t r a c t

Article history: Abatement of fluorinated compounds (FCs) used in semiconductor and display industries has received
Received 19 November 2014 an attention due to the increasingly stricter regulation on their emission. We have developed a 2.45 GHz
Received in revised form 6 March 2015 microwave plasma torch with reverse vortex reactor (RVR). In order to design a reverse vortex plasma
Accepted 7 March 2015
reactor, we calculated a volume fraction and temperature distribution of discharge gas and waste gas in
Available online 21 March 2015
RVR by ANSYS CFX of computational fluid dynamics (CFD) simulation code. Abatement experiments have
been performed with respect to SF6 , NF3 by varying plasma power and N2 flow rates, and FCs concentra-
Keywords:
tion. Detailed experiments were conducted on the abatement of NF3 and SF6 in terms of destruction and
Fluornated compounds abatement
Microwave plasma torch removal efficiency (DRE) using Fourier transform infrared (FTIR). The DRE of 99.9% for NF3 was achieved
Reverse vortex reactor without an additive gas at the N2 flow rate of 150 liter per minute (L/min) by applying a microwave
power of 6 kW with RVR. Also, a DRE of SF6 was 99.99% at the N2 flow rate of 60 L/min using an applied
microwave power of 6 kW. The performance of reverse vortex reactor increased about 43% of NF3 and 29%
of SF6 abatements results definition by decomposition energy per liter more than conventional vortex
reactor.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
Several fluorinated compounds (FCs) used in electronic man-
ufacturing, including NF3 , SF6 , perfluorocarbons (PFCs), and
hydrofluorocarbons (HFCs) are extremely potent greenhouse gases
[1]. Gases in the atmosphere can contribute to the greenhouse effect
both directly and indirectly. Direct effect occurs when the gas itself
is a greenhouse gas like water vapor, carbon dioxide (CO2 ), methane
(CH4 ) and nitrous oxide (N2 O). Indirect radiative forcing occurs
when chemical transformations of the original gas produce other
greenhouse gases, when a gas influences the atmospheric lifetimes
of other gases, and/or when a gas affects atmospheric processes
that alter the radiative balance of the earth. The intergovernmental
∗ Corresponding author. Tel.: +82 42 879 5912.
E-mail address: ychong@nfri.re.kr (Y.C. Hong).
panel on climate change (IPCC) ranks sulfur hexa-fluoride (SF6 ) and
nitrogen tri-fluoride (NF3 ) as having the highest global warming
potential (GWP) among all greenhouse gases; 22,800 and 17,200
times greater than CO2 , which is based on the first 100 years in the
atmosphere (GWP100 ) [2]. Its atmospheric lifetimes are estimated
to be 740 of and 3200 years of NF3 and SF6 .
The strategies evaluated by industry for reducing FCs emissions
include applying substitute chemicals, recovery and recycle, pro-
cess optimization, and abatement technologies. Although the first
three options may be more desirable from and environmental stand
point, there are many obstacles to their implementation because
of the complex nature of the semiconductor manufacturing pro-
cess. The PFCs abatement technologies were studied using catalyst
[3], combustion [4], superheated steam, solid alkaline reaction [5],
non-thermal plasma [6,7], and thermal plasma [8,9]. Currently,
atmospheric pressure plasmas are studied with growing interest,
because it provides a more economic and convenient alternative to

http://dx.doi.org/10.1016/j.jhazmat.2015.03.016
0304-3894/© 2015 Elsevier B.V. All rights reserved.
42 J.H. Kim et al. / Journal of Hazardous Materials 294 (2015) 41–46

Fig. 1. Schematics of the reverse vortex reactor (a) and details (b) for CFD simulation.

low pressure plasma technology. The high energy density and tem-
peratures associated with thermal plasmas and the corresponding
fast reaction times offer the potential of large throughputs in a small
reactor [8,9]. There are several kinds of plasma torches, and a high
frequency capacitive torch. The direct current (DC) arc torch is oper-
ated by the DC electric field between the two electrodes, which
must be replaced often due to their limited lifetime. The induc-
tive coupled plasma (ICP) torch is inefficient devices with typical
thermal efficiency in the range of 40–50% and difficult to ignite
a discharge [10]. A Microwave plasma torch has several advan-
tages of electrodeless discharge, thermal and large volume plasma.
Previously, an atmospheric pressure microwave plasma torch as a
tool for FCs abatement was studied but capability of flow rates is
low about 10–15 liters per minute (L/min), waste gases included
Si particle bring about microwave transfer have to pass through
the discharge tube and heat loss of radiation occurred from the
reactor wall. In order to apply the industry, 3 problems of capabil-
ity, microwave transfer and heat loss would be improved. In this
study, we designed a new concept of reverse vortex plasma reactor
to increase a flow rate of waste gas, not to pass through a discharge
tube, to decrease a heat loss from outer wall of the reactor. We have
investigated the abatement of NF3 and SF6 gases which are mixed
in nitrogen gas as an emitted gas of etching process. The additive
gases, such as H2 and O2 , were not injected into the system to attain
the developed decomposition of NF3 and SF6 . It has been expected
that we will get better destruction and removal efficiency (DRE) if
diluted gas using in the semi-conductor industry instead of pure
SF6 is treated by thermal plasma.

Fig. 2. Computational results for the reverse vortex reactor model in Fig. 1. (a) Fluid streamlines in the reactor, (b) volume fraction distributions in the reactor, (c) temperature
distributions in the reactor.
 J.H. Kim et al. / Journal of Hazardous Materials 294 (2015) 41–46 43

2. Reverse vortex reactor design

We designed a plasma reactor for increasing the flow rate of

FCs gases mixed in N2 gases, and decreasing a heat loss from the
out of the reactor wall required for industrial use. The reverse vor-
tex reactor (RVR) is better gas residence time and heat losses than
conventional vortex reactor [11,12]. To better understand the fluid
behavior and temperature distribution in the RVR, we modeled the
flows by using CFD code of ANSYS CFX software. In order to simu-
late a RVR, we defined boundary conditions of input power, swirl
input (discharge gas) flow rate, reverse vortex (waste gas) input
flow rate and inlet gas temperature.
Fig. 1 shows the geometry (a) and details for simulation (b) of
RVR. Basically, 2.45 GHz microwave plasma torch equipped with
quartz tube to serve as a discharge tube and with feeders. The diam-
eter of D2 is defined 26 mm, because the inner diameter of quartz
is 26 mm, also, the diameter of D1 is 40 mm. The distance from exit
quartz tube to the inner cylinder of “L” is parameter of simulation
conditions by 5, 10, 15 mm.
The simulation results on the stream lines (a), volume fraction
(b), and temperature distribution contour (c) in a reactor shown in
Fig. 2. Fig. 2(a) is presented fluid stream lines that shows a behav-
ior of fluid 1 (discharge gas), fluid 2 (waste gas) and mixing in a
reactor. The simulation results of volume fraction and temperature
graphically were shown in Fig. 3. In the result of Fig. 3(a), when “L”
is 5 mm, volume fraction of fluid 1 is highly distributed all over the
radial direction region more than 10 mm and 15 mm. In the results
of fluid temperature, when “L” is 5 mm, fluid temperature is higher
than other conditions at the center of inner reactor, however, is
lower than other conditions near of the reactor wall. We selected
“L” value of 5 mm for design and fabrication as a simulation results.

3. Experiments Fig. 3. Computational results for the temperature (a) and volume fractions (b) across
radii at selected elevations in the reverse vortex reactor models of Fig. 1.
The FCs abatement system used in this study is an atmospheric
pressure 2.45 GHz microwave plasma torch. As shown schemati-
cally in Fig. 4, the main parts of the system consist of the microwave
plasma torch and the plasma reactor. Microwave radiation is

Fig. 4. Schematic and picture showing an experimental setup for the microwave plasma torch abatement of NF3 and SF6 with reverse vortex reactor.
44 J.H. Kim et al. / Journal of Hazardous Materials 294 (2015) 41–46

Fig. 5. Reactor wall temperature of the CVR and RVR.

generated by using a 2.45 GHz magnetron and it pass through a

directional coupler, a three-stub tuner, and a tapered waveguide,
where it enters a discharge tube made of fused quartz. The cen-
ter axis of the quartz discharge tube, with an outer diameter of
approximately 30 mm, is located one-quarter wavelength from the
short end of the waveguide and is perpendicular to the wide waveg-
uide walls. The electric field induced by the microwave radiation
in the quartz tube can be maximized by adjusting the three-stub
tuner to reduce the reflected power to less than 1% of the forward
power. In our FCs abatement experiments using a compressed air
supply system, liquid nitrogen gas supply system for the plasma
discharge and making a simulation gas. The FCs gases of NF3 and
SF6 were injected from gas barrel and controlled gas flow rate by a
mass flow controller (MFC). The plasma generated inside the dis-
charge tube is stabilized by input of swirling air and meet the waste
gas mixed by nitrogen and SF6 or NF3 . Mixed gases are ionized by
electron from the air plasmas. In order to increase a gas residence
Fig. 6. FTIR spectra illustrating abatement of NF3 by making use of a microwave
time, conventional reactor was extended after RVR. Fig. 4(a) was plasma torch ver. input power (a) and N2 flow rate (b).
shown a schematically and picture of FCs abatement system using
by microwave plasma torch. A picture of microwave plasma torch
with RVR shown in Fig. 4(b) dissociated to NFx (x = 0, 1, 2) radicals by electron impact processes,
In order to analysis the heat loss of the reactor, we measured the and the reaction [13] is expressed as
temperature of the reactor wall. Fig. 5 shows measurement results NF3 → NFx + Fy (k = 6.6 × 10−8 e−24160/T cm3 /s), as (1)
of the reactor wall of the CVR and RVR. The differences of tem-
perature between CVR and RVR are from 115 ◦ C to 250 ◦ C, because where y is 1 or 2 and T is in range of 1180–1800 K. Compressed
outer reverse vortex flows protect heat transfer and heated reverse air was injected as a swirl gas of the microwave plasma torch for
vortex flows enter into the inner reaction chamber. oxidation with NF3 radicals. When oxygen as included air is used
As shown in Fig. 4, FTIR diagnostic was used in the system for to abate NF3 , the desired reaction pathway of oxygen is to oxidize
an in-depth understanding of the plasma destruction process. A
FTIR measurement is widely used as a powerful method to mon-
itor PFC emission stacks for destruction and removal efficiency
(DRE). The FTIR used in this experiment is the VERTEX 80, made
by BRUKER Co., in the Germany, for quantitatively identifying con-
centrations of outlet streams before and after plasma discharge. The
FTIR, scanning range of 10–5000 cm−1 at 0.06 cm−1 spectral reso-
lution and visible spectral range demonstrate a resolving power of
better than 300,000:1, is equipped with an IR window. An extrac-
tive sampling method was employed to sample plasma byproduct
gases exhausted from the exhaust line. In addition to Fig. 4, HF was
treated through the wet scrubber. Although NO2 , NO, SO2 , SO2 F2 ,
and SOF2 are well known as water-soluble products, these products
of a small amount, not solved in the wet scrubber, are found in our
experiments.
In the experiment of NF3 or SF6 abatement, air injected into
the microwave plasma torch and NF3 and SF6 were mixed with N2
in the gas-mixing vessel and tangentially injected into the reverse Fig. 7. FTIR spectra illustrating abatement of SF6 by making use of a microwave
vortex reactor. The mixed gases are indirectly ionized, attached, or plasma torch. The operating power of the plasma torch was 6 kW.
 J.H. Kim et al. / Journal of Hazardous Materials 294 (2015) 41–46 45

the nitrogen in NF3 to Nx Oy . These oxygen atoms react with the

radicals. The chemical reaction equations [14] are

O + NF2 → NF + OF(k = 10−12 cm3 /s) (2)

−12 3
O + NF → NO + F(k = 10 cm /s) (3)
−11 3
O + OF → O2 + F(k = 5 × 10 cm /s). (4)

Based on Eqs. (2)–(4), the final byproducts are nitrogen monox-

ide and fluorine at the downstream of the reactor. When H2 O is
also used to abate NF3 , electron impact dissociation produces O, H,
and OH radicals. O and H radicals are precursors for FC remediation.
The OH radical can also remediate FCs, both by oxidizing NFx and
by preventing its re-association to make NF3 .
In the addition of NF3 abatement, oxygen atoms also react
with SFx (x = 1 − 5) radicals produced by electron impact processes,
creating additional SO2 or SO and forming HF, SOF2 and SO2 F2
molecules by F2 reactions downstream of the plasma. The chemical
reaction equations [15–18] are Fig. 8. Plots of NF3 and SF6 DREs vs N2 flow rates applying the plasma power of
6 kW.
O + SF5 → SOF4 + F(k = 2 × 10−12 cm3 /s) (5)

O + SF4 → SOF4 (k = 2 × 10 −14

cm /s) 3
(6) wavenumber 941 and 984 cm−1 on the horizontal line represents
the signature of the SF6 concentration. From Fig. 8, it seems that
O + SF2 → SOF2 (k = 1.08 × 10−10 cm3 /s) (7) the abatement of SF6 is more difficult than that of NF3 . Fig. 8 shows
−11 3 a DREs of NF3 and SF6 versus energy density.
O + SF2 → SOF + F(k = 7.63 × 10 cm /s) (8)
−10 3
Fig. 9 shows comparison results of NF3 abatement about RVR
O + SF → SOF + F(k = 1.7 × 10 cm /s) (9) and CVR. DREs result of the CVR, we referred to previous work [8]
−11 3 for DREs of NF3 by applying the microwave power of 1.4 kW at flow
O + SOF → SOF4 + Products(k = 2 × 10 cm /s) (10)
rate of 20 L/min N2 . Also, SF6 DREs of CVR in Fig. 10 were referred
4. Results and discussion

Quantitative gas-phase Fourier transform infrared (FTIR) spec-

troscopy was employed to identify the reduction of NF3 , SF6
concentration and concentration changes of the plasma by products
along with time before and after microwave discharge. DRE repre-
sents the percentage of FCs that has been destroyed. The definition
of DRE is
 (S 
− Safter )
before
DRE(%) = × 100 (11)
(Sbefore )
where, Sbefore and Safter are the main peak areas of the NF3 and
SF6 before and after discharge of the microwave plasma torch.
Table 1 shows the results for the determination of the DREs
of the NF3 and SF6 , depending on the applied microwave power.
Experiments were carried out to investigate the destruction of NF3
and SF6 for recipes 1 and 2, and 3–6, respectively, using compressed
air and oxygen as swirl gases, ethylene as an additive gas, and dif- Fig. 9. DREs plotted as a function of the applied energy density for conventional
ferent applied microwave powers. Fig. 6 shows abatement results vortex reactor [8] and reverse vortex reactor on the NF3 .
of NF3 FTIR spectra by the plasma power (a) and N2 flow rate (b). In
the Fig. 6(a), the mixture of 150 L/min N2 and 0.15 L/min NF3 , and
60 L/min compressed air as a swirl gas were injected by applying
the microwave power of 0–6 kW. Although the microwave plasma
torch in gas discharge provides high plasma density and high gas
temperature in inexpensive ways, the plasma volume and tempera-
ture of the microwave plasma torch decrease drastically outside the
discharge tube. When the plasma power is increased, the plasma
temperature and the length are increased. So the reaction proba-
bility and gas residence time increased DRE per unit flow rate. In
case of Fig. 6(b), microwave power of 6 kW applied to the torch
at mixture of 150, 200, 250 L/min N2 with NF3 of 1000 ppm. The
wavenumber 910 and 1032 cm−1 on the horizontal line represents
the signature of the NF3 concentration. As expected, the plasma
torch flame, generated by 6 kW microwave power destroyed most
of the NF3 contaminants (DRE > 99.9%) at 150 L/min N2 . SF6 FTIR
spectra were shown in Fig. 7 along with N2 flow rates. In the Fig. 8,
microwave power of 6 kW applied to the torch at mixture of 60, 80, Fig. 10. DREs plotted as a function of the applied energy density for conventional
100, 120 L/min N2 , and SF6 of 0.12, 0.16, 0.20, and 0.24 L/min. The vortex reactor [9] and reverse vortex reactor on the SF6 .
46 J.H. Kim et al. / Journal of Hazardous Materials 294 (2015) 41–46
Table 1
The destruction and removal of NF3 and SF6 .
No. NF3 (L/min) SF6 (L/min) N2 (L/min)
1 2
2 3
3 0.15 150
4 0
5 99.9
6 0.2 200
7 0.25 250
8 0.12 60
9 0.16 80
0
10 0.2 100
11 0.24 120
from previous work [9]. The comparison results between CVR and
RVR, we obtain improved results that destruction energy density
decreased to 1800 and 2400 J/L over 99.9% of NF3 and SF6 DREs.
5. Conclusions
In this study, we present the improvement and advantages of
using an atmospheric pressure microwave plasma torch for the
destruction and removal of chemically stable NF3 and SF6 emit-
ting from semiconductor fabrication system with reverse vortex
plasma reactor. In order to improve a DRE and capability of flow
rate, we developed a reverse vortex plasma reactor base on the CFD
simulation results of volume fraction and temperature with ANSYS
CFX. The capability of discharge gases was increased over the 3–5
times because of reverse vortex plasma reactor and then, capac-
ity of abating flow rate were increased also about 300 L/min. For
NF3 abatement, we achieved a DRE of more than 99.99% in mixture
0.15 L/min NF3 and 150 L/min N2 by applying the microwave power
6 kW. In case of SF6 , DRE was 99.99% in mixture of 0.18 L/min SF6
and 60 L/min N2 by applying the microwave power of 6 kW. It seems
that the abatement of SF6 is more difficult than that of NF3 . More-
over, the waste gases not pass through a discharge tubes, expected
interruption of microwave transmission cause by particle depo-
sition in the discharge tube is avoidable. The destruction energy
densities of NF3 and SF6 are each 2400 and 6000 J/L over 99.9% DREs
by RVR because of differences of heat losses from CVR and RVR wall.
The performance of destruction energy densities were improved
performance of 43% of NF3 and 29% of SF6 compared with CVR.
The aforementioned characteristics make the microwave plasma
torch attractive for the destruction and removal of chemically sta-
ble gases emitted by the semiconductor industry.
Air (L/min) Plasma power (kW) DRE (%)
33.2
78.3
4 98.3
5 99.1
60 98.4
84.2
6 99.99
86.3
82.4
64.2
Acknowledgments
This work was supported by R&D Program of “Plasma Con-
vergence & Fundamental Research” through the National Fusion
Research Institute of Korea (NFRI) funded by the Government funds
and in part by the Degree & Research Center Program of the Korea
Research Council of Fundamental Science and Technology.
References
[1] J.-H. Cheng, C. Scoot Bartos, W.M. Lee, S.-N. Li, J. Lu, Int. J. Greenhouse Gas
Control 17 (2013) 106.
[2] M.B. Chang, J.S. Chang, Ind. Eng. Chem. Res. 45 (2006) 4101.
[3] M.B. Chang, H.M. Lee, Catal. Today 89 (2004) 109.
[4] L. Beu, P.T. Brown. Proceedings of the IEEE/CPMT International Electronic
Manufacturing Technology Symposium, 1998, 277.
[5] M.C. Lee, W. Choi, Environ. Sci. Technol. 36 (2002) 1367.
[6] D.H. Kim, Y.S. Mok, J. Korean Phys. Soc. 59 (2011) 3437.
[7] M. Sanjeeva Gandhi, Y.S. Mok, J. Environ. Sci. 24 (2012) 1234.
[8] D.H. Shin, Y.C. Hong, S.C. Cho, H.S. Uhm, Phys. Plasmas 13 (2006) 114,504.
[9] Y.C. Hong, H.S. Uhm, B.J. Chun, S.K. Lee, S.K. Hwang, D.S. Kim, Phys. Plasmas 13
(2006) 33,508.
[10] Y.C. Hong, S.C. Cho, C.U. Bang, D.H. Shin, J.H. Kim, H.S. Uhm, W.J. Yi, Appl. Phys.
Lett. 88 (2006) 201,502.
[11] J.H. Kim, S.H. Ma, C.H. Cho, Y.C. Hong, J.Y. Ahn, J. Korean Phys. Soc. 64 (2014)
238.
[12] Y.C. Hong, J.H. Kim, S.H. Ma, C.H. Cho, IEEE Trans. Plasma Sci. 64 (2014)
1982.
[13] D.L. Baulch, J. Duxbury, S.J. Grant, D.C. Montaque, J. Phys. Chem. Ref. Data.
Suppl. 42 (2014) 1.
[14] K.E. Greenbug, J.T.L. Verdeyen, J. Appl. Phys. 57 (1985) 1596.
[15] C.J. Ultee, Chem. Phys. Lett. 46 (1997) 366.
[16] K.R. Ryan, I.C. Plumb, Plasma Chem. Plasma Process. 8 (1988) 281.
[17] J.T. Herron, Int. J. Chem. Kinet. 19 (1987) 129.
[18] I.C. Plumb, K.R. Ryan, Plasma Chem. Plasma Process. 6 (1986) 247.