22nd International Symposium on Plasma Chemistry
July 5-10, 2015; Antwerp, Belgium
Pulsed dry methane deforming in DBD and catalyst hybrid reaction
T. Nozaki1, S. Kameshima1, K. Tamura1, Y. Ishibashi1 and K. Okazaki2
1
2
Department of Mechanical Sciences and Engineering, Tokyo Tech., Tokyo, Japan
Department of Mechanical and Control Engineering, Tokyo Tech., Tokyo, Japan
Abstract: Pulsed dry methane reforming (DMR) in dielectric barrier discharge (DBD) and
Ni/Al 2 O 3 catalyst hybrid reaction was investigated, aiming for efficient conversion of
greenhouse gas (CH 4 , CO 2 ) into syngas (H 2 , CO) at low temperature. Pulsed transient
analysis reveals that CH 4 dehydrogenation and subsequent reverse water-gas-shift reaction
is sufficiently fast with and without DBD. In contrast, carbon removal reaction is promoted
clearly by DBD hybridization. In order to gain further insight into complex plasma-surface
interaction, optical emission spectroscopy of CO Ångström system and C 2 high pressure
Swan system was performed.
Keywords: Plasma catalysis, Optical emission spectroscopy, Hydrogen, Greenhouse gas.
1. Introduction
In recent year, power-to-gas conversion technology is
highlighted where water electrolysis driven by renewable
electricity is combined with catalytic conversion of CO 2
[1]. CO 2 is converted into CH 4 with the renewable H 2 ,
then distributed through existing gas grid for widespread
use. With a similar concept, we propose greenhouse gas
conversion (CH 4 and CO 2 ) using non-thermal plasma
enhanced catalytic reaction. This technology is known as
dry methane reforming (DMR) and a mixture of CO and
H 2 is synthesized. Analogous to electrolysis of water
splitting, renewable electricity is converted into chemical
energy via non-thermal plasma assisted endothermic
reactions. Syngas is then converted preferably into carbon
containing liquid fuels whose energy density is 10−100
times greater than that of solid state secondary batteries.
As a result, transport and storage capability of renewable
energy is greatly improved and CO 2 utilization is realized
simultaneously. Currently, electrochemical reaction is
dominantly studied for power-to-gas conversion [1,2].
Besides, non-thermal plasma provides additional energy
and material conversion pathways, contributing to an
extended carbon recycling network and fuel flexibility [24].
For this purpose, pulsed dry methane reforming (CH 4 +
CO 2 = 2CO + 2H 2 ) in Ni/Al 2 O 3 catalysts and dielectric
barrier discharge (DBD) hybrid reaction was investigated.
CH 4 was injected for short duration (1 min), while CO 2
plasma was generated alternately to remove solid carbon
precipitated onto the catalysts. Examining time dependent
consumption of CH 4 and CO 2 and the evolution of CO
and H 2 profiles, rate-controlling step and related kinetics
were investigated. To obtain a better insight into complex
plasma-surface interaction, optical emission spectroscopy,
catalyst bed temperature analysis, and time dependent
changes of those quantities are also investigated. We
found that hybrid reforming enables better carbon
deposition control, allowing continuous operation without
O-12-4
serious coke formation. Reforming kinetics, coke
formation mechanism and synergistic effect brought by
DBD are comprehensively discussed.
2. Experimental
Fig. 1 shows an overview of the catalyst bed, optical
emission and temperature distribution. Optical emission
spectroscopy was employed through the slit at locations a
(inlet) and b (outlet) as in Fig. 1(a). The air breaks down
at the sharp edge of the ground electrode as pointed by
arrows in Fig. 1(b): N 2 second positive system is visible,
but this event occurs outside of reactor.
Fig. 1. (a) Overview of catalyst bed, (b) Optical emission from
DBD. (c) Thermograph showing temperature distribution of the
bed.
Detailed reactor configuration is provided elsewhere
[5]. Briefly, a quartz tube with 20 mm inner diameter was
used for the packed-bed DBD reactor. Catalyst pellets
(ISOP, 12 wt.% Ni/Al 2 O 3 and φ 3 mm; Süd-Chemie
Catalysts Japan, Inc.) were packed over 20 mm length and
both sides were supported by non-catalytic Al 2 O 3 pellets
(φ 3 mm) as shown in Fig. 1(a). Catalysts were reduced
for 90 min in H 2 /N 2 = 50/450 cm3/min at 600 °C prior to
the experiment. Stainless steel rod with 3 mm diameter
was inserted into the center of the quartz tube as a high
voltage electrode. Ground electrode has a slit (10 mm
1
width, see Fig. 1) and the temperature distribution of the
catalyst bed was measured by thermography (TH5140;
NEC San-ei Instruments Ltd.). A high voltage power
source was used to generate DBD (Logy Electric LHV13AC; 13–14 kHz). Discharge power was measured by
Lissajous diagram with a capacitor (0.036 µF). DBD is
extinguished when the solid carbon is deposited on the
pellet surface. Therefore, pressure was evacuated till 5
kPa so that DBD is generated independently of carbon
deposition. Output gas flows into a cold trap (ca. −40 °C)
and H 2 O was removed. After that, CO 2 , CH 4 , CO and H 2
were monitored online by a quadruple mass spectrometer
(QMS, Prisma-100; Pfeiffer Vacuum GmbH).
3. Pulsed dry methane reforming with DBD
Pulsed reforming was carried out for 56 minutes (4 min
per cycle × 14 cycles) and time dependent change in gas
composition was investigated. Catalysts were pre-heated
in N 2 flow (1000 cm3/min) until the catalyst bed
temperature reached steady state. This temperature is
defined as initial temperature as shown in Fig. 2. CO 2
flow rate was kept constant, whereas CH 4 was supplied
intermittently at constant interval. Reforming conditions
are summarized in Table 1.
Figures 2(a) and (b) highlight the last three cycles of the
reaction with and without DBD. In Fig. 2(a), at the
moment of CH 4 injection, H 2 and CO are produced
abruptly. Sharp increase in H 2 implies that H 2 is released
by stepwise dehydrogenation of adsorbed CH 4 * and
surface recombination of H*. Adsorbed H* is also
consumed by CO 2 to form CO, which is known as the
reverse water-gas-shift (R-WGS) reaction. Production of
CO is slower than that of H 2 , thereby CO signal is
slightly delayed in transient phase. A similar trend was
observed in transient analysis of catalytic DMR [6].
4. Coke formation and de-coking by CO 2 -DBD
After abrupt increase in H 2 concentration, it decreases
gradually because catalyst temperature decreases by the
endothermic reaction. After 1 min operation of DMR,
CH 4 flow was turned off and CO 2 plasma was generated
for 3 min. According to the Boudouard equilibrium at
given catalyst temperature, solid carbon is oxidized and
removed as CO:
CO 2 + C or NiC → CO + CO
R1
CO concentration exhibits peak as indicated by arrow A
in Fig. 2(a), then turns to decrease gradually as solid
carbon consumed. Oxidation of solid carbon is completed
in about 1.2 min (arrow B); Correspondingly, CO 2
concentration approaches constant value. Solid carbon
was almost fully oxidized by CO 2 which is also
confirmed by Raman analysis of residual carbonaceous
material (data is not shown).
Fig. 2. Time dependent change in gas composition and catalyst
bed temperature. (a) hybrid reaction and (b) thermal reaction.
Conditions: see Table 1.
These observations derive three important hypotheses:
First, CO 2 cannot be dissociated on Ni-only surface [7].
Adsorbed H* or C* is necessary to break C=O bond of
CO 2 . Second, solid carbon is not produced from CO.
Third, solid carbon is therefore produced via stepwise
dehydrogenation of adsorbed CH 4 *. Figure 3
schematically depicts carbon deposition mechanism. First,
carbon is precipitated via CH 4 dehydrogenation (Fig.
3(a)). Carbon concentration near catalyst surface is high
and carbon diffuse over catalyst, while producing nickel
carbide (NiC) (Fig. 3(b)), which is thermodynamically
favored in the carbon rich conditions [8]. When carbon is
excessively deposited, filamentary carbon grows (Fig.
3(d)). Carbon precipitation is terminated when CH 4 flow
is turned off. Subsequently, solid carbon is oxidized by
adsorbed CO 2 * (Fig. 3(c)). The surface carbon is readily
oxidized; however, carbon near the center must diffuse
towards the surface and then oxidized. Carbon diffusion
through nickel or nickel carbide becomes the ratecontrolling step. Therefore, carbon removal reaction (R1)
is slow: Correspondingly, CO concentration varies
moderately in the de-coking phase (Fig. 2).
Table 1. Experimental conditions. (1) Standard conditions
Reforming
(0 < t < τ)
De-coking
(τ < t < T)
2
GHSV(1)
(h-1)
Reaction
Time (ms)
SEI
(eV/molecule)
9260
6.4
1.07
27800
19.2
3.22
τ (min)
T
(min)
1.0
4.0
CH4/CO2
(cm3/min)
Power
(W)
P (kPa)
70
5.0
666/333
0/333
O-12-4
concentration decreases; however, it turns to increase
because Boudouard reaction play role (R1). C 2 emission
is fairly weak in the reforming phase and abruptly
increased at t = 1.25 min, which is slightly after CH 4 flow
was turned off. After that, C 2 emission maintained a
certain level, then rather abruptly weakened at t = 1.8 min.
Fig. 3. Mechanism of coking and de-coking.
5. Optical emission spectroscopy: overview
Figure 4 shows optical emission spectrum observed at
the reactor outlet (point b in Fig. 1(a)). Spectrometer
(USB4000; Ocean Optics, Inc.) was synchronized with
pulsed CH 4 supply and the spectrum shown in Fig. 4(a)
was recorded at 5 seconds after CH 4 flow was turned off
(beginning of de-coking phase). Spectrum consists of the
Ångström and the third positive system of CO which is
overlapped with the N 2 second positive system. N 2
emission is caused by the gas breakdown outside of the
reactor (Fig. 1(b)). CH emission is commonly observed in
packed-bed DBD when CH 4 is admixed [9,10]; however,
it is missing in this study. Emission from OH was not
observed, although H 2 O is produced via reverse watergas-shift reaction. Adsorbed OH* play important role on
surface reaction, but gas phase OH was undetectable.
Moreover, CO 2 + emission (290 nm) is missing in our
study which is often reported in CO 2 /CH 4 DBD [9,10]. In
contrast, strong emission at 589.0 nm was uniquely
observed in this study. Fine structure was further analyzed
by 750 mm focal length spectrometer equipped with 1200
grooves/mm grating (SP2750; ACTON Research),
showing two lines at 589.0 nm and 589.2 nm (Fig. 4(b)).
Emission near 589.0 nm is observed in CO plasma [11],
which is explained as high pressure Swan system of C 2
[12,13]. Figure 4(b) exhibits atomic-like narrow peaks
whose line width is as narrow as fine structure of
rotational branch of CO Ångström system. General
appearance of C 2 high pressure Swan system is double
headed narrow peaks.
6. Plasma-surface interaction: de-coking mechanism
Time dependent change of the line intensity of CO
(519.8 nm) and C 2 high pressure Swan system are
presented in Fig. 5 together with mass spectrum profile of
CO, CO 2 , H 2 and CH 4 . Emission spectrum was obtained
at the reactor outlet and the intensity is normalized based
on the intensity of C 2 peak at t = 1.25 min. It is evident
that intensity of CO Ångström at the outlet is well
correlated with CO mass spectrum profile over the cycle.
When CH 4 flow is turned off, CO 2 concentration
increases because R-WGS reaction no longer occurs due
to an absence of adsorbed H*. Correspondingly, CO
O-12-4
Fig. 4. Emission spectrum
recorded at reactor outlet. (a)
Snapshot spectrum observed
at t = 1.25 min (see Fig. 5).
●, N 2 second positive
system; ▲, CO third
positive system; ○, not
identified. (b) High
resolution spectrum of C 2
high pressure Swan system.
Emission from C 2 is initiated by the production of
atomic carbon. Atomic carbon is produced by
vibrationally excited CO and electrically excited
metastable state CO( 𝑎3Π) [12,13]:
CO(v) + CO(w) = C + CO2
CO(v) + CO(w) = CO(3aΠ) + CO
CO + e = CO( 3aΠ) + e
CO( 3aΠ) + CO = CO2 + C
R2
R3
R4
R5
C + C + M = C2 (d 3Πg , v = 6)
C + CO + M = C2 O + M
C + C2 O = C2 �d 3Πg , v = 6� + CO
R6
R7
R8
CO is consumed by R2 through R5 to form carbon
atoms. Carbon atoms recombine to form C 2 . Diatomic
carbon (C 2 ) is also produced via intermediate species
(C 2 O). Atomic carbon appearing in R2−R8 is most likely
in the ground state (C(3P)).
The triplet 𝐶2 �d 3Π𝑔 , 𝑣 = 6� level lies at 3.71 eV [14]
and selectively populated via collision-induced
vibrational energy transfer. Life time of this state is
relatively short, and depopulated to the ground state with
optical emission, known as high pressure Swan system.
3
Solid carbon is concentrated near the catalyst surface as
shown in Fig. 3(a)-(b). Therefore, production of CO via
R1 is maximized in the initial stage of de-coking phase.
Production of C and C 2 via R2−R8 is also maximized,
resulting in sharp increase in C 2 emission. Surface carbon
is rapidly consumed and the carbon removal reaction is
substantially slowed because carbon diffusion across
catalyst particle becomes the rate-determining step.
Therefore, C 2 emission is closely correlated with surface
reaction. Further understanding of gas phase CO
activation and resulting C 2 emission mechanism would
provide better insight into plasma-surface interaction.
understanding of the complex interaction of non-thermal
plasma and catalyst surface is needed for better carbon
deposition control with improved material and energy
conversion efficiency.
8.References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
Fig. 5. Top: Emission intensity of CO (518.3 nm) (solid blue
line) and C 2 (dotted black line). Bottom: Mass spectrum profile
for one cycle pulsed reforming. Emission profile is normalized
by emission intensity of C 2 peak at 1.25 min.
[14]
S Schiebahn, T Grube, M Robinius, L Zhao, A Otto et al.,
“Power to Gas” in Transition to Renewable Energy
Systems, Ed. D. Stolten and V. Scherer: Wiley-VCH
(2013) pp. 813-847.
M. Tahir, N.S. Amin, Renew. & Sustain. Energy Rev., 25
(2013) 560–579.
T. Nozaki, K. Okazaki, Green Proc. & Synthesis, 6 (2012)
517−523.
G. Petitpas, J.-D. Rollier, A. Darmon, J. GonzalezAguilar, R. Metkemeijer, L. Fulcheri, Int. J. Hydrogen
Energy, 32 (2007) 2848−2867.
T. Nozaki, H. Tsukijihara, K. Okazaki, Energy and Fuels,
20 (2006) 339−345.
T. Osaki, T. Horiuchi, K. Suzuki, T. Mori, J. Chem. Soc.
Faraday Trans., 92 (1996) 1627−1631.
J. Guo, J. Gao, B. Chen, Z. Hou, J. Fei, H. Lou, X. Zheng,
Int. J. Hydro. Energy, 34 (2009) 8905−8911.
G.A. Jablomski, F.W. Geurts, A. Sacco, Jr., Carbon, 30
(1992) 87−98.
M. Kraus, W. Egli, K. Haffner, B. Eliasson, U.
Kogelschatz, A. Wokaun, Phys. Chem. Chem. Phys., 4
(2002) 668−675.
X. Tu, J.C. Whitehead, Appl. Catal. B: Environ., 125
(2012) 439−448.
R. Geiger and D. Staack, J. Phys. D: Appl. Phys., 44
(2011) 274005(13 pp.).
C.E. Little and P.G, Browne, J. Phys. B, Atom. Mol. Opt.
Phys., 22 (1989) 1269−1283.
C.E. Litle and P.G. Browne, Chem. Phys. Lett., 134 (1987)
560−564.
H.L. Wallaart, B. Piar, M.Y. Perrin, J.P. Martin, Chem.
Phys., 196 (1995) 149−169.
7. Conclusion
Pulsed dry methane reforming was investigated using
catalytic packed-bed DBD reactor. Pulsed reforming
enables continuous operation without serious coking
problem. Dehydrogenation of CH 4 on Ni catalysts and
subsequent R-WGS reaction are sufficiently fast with and
without DBD. Solid carbon is produced most likely by
CH 4 dehydrogenation and those carbon forms nickel
carbide which is thermodynamically favored in the given
conditions. In contrast, oxidation of solid carbon by CO 2
is remarkably slower than reforming reaction because
carbon must diffuse through the nickel or nickel carbide
to be removed by adsorbed CO 2 .
C 2 high pressure Swan system was uniquely observed
during the de-coking phase. Excited C 2 is produced by
vibrationally excited CO which is derived from solid
carbon and adsorbed CO 2 (R1). It implies that C 2
emission is closely correlated with de-coking reaction
which occurs on the catalyst surface, providing better
insight into plasma-surface interaction. Further
4
O-12-4