Plasma-based Dry Reforming in a Dielectric Barrier Discharge:
a Computational Study
R. Snoeckx, R. Aerts, A. Bogaerts
Research group PLASMANT, Department of Chemistry, University of Antwerp, Antwerp, Belgium
Abstract: We present a computational study of a dielectric barrier discharge used for the
conversion of CH4 and CO2 into value-added chemicals, i.e., dry reforming of methane. A
zero-dimensional chemical kinetics model is applied to study the plasma chemistry and the
chemical conversion pathways. The calculated conversions, selectivities and energy
efficiency are compared with experimental values, and reasonable agreement is achieved.
Keywords: Dielectric barrier discharge, Numerical simulations, Methane, Carbon dioxide.
1.
Introduction
In many respects methane is an attractive fuel for
heating and electrical power generation. However, this
makes methane an underutilized source for the production
of useful valuable chemicals and fuels, such as hydrogen
gas, higher hydrocarbons, syngas, methanol and
formaldehyde. Both methane itself and carbon dioxide—
derived from oxidizing methane—are greenhouse gases.
The conversion of these gases into value-added
chemicals or fuels is considered as one of the main
challenges for the 21st century.[1], [2] This way, the
(greenhouse) gases can constitute an alternative for
petroleum, which will become less available, and
therefore more expensive. Moreover, this conversion
process can be considered to fit in the revolutionary
“cradle-to-cradle” concept,[3] because the greenhouse
gases (waste) can be converted into raw materials for the
chemical industry (new feedstock).
The conversion process of CO2 and CH4 is, however,
not straightforward. The indirect synthesis routes for the
utilization of CH4 require syngas as an intermediate step.
The most important processes for the intermediate syngas
step are steam methane reforming (SMR), dry reforming
of methane (DRM) and partial oxidation of methane
(POX). These processes are often followed by a methanol
or Fischer-Tropsch synthesis to obtain the desired
products.[4] The latter methods are, however,
characterized by low overall yields and they require a
high energy input.[5] Direct (thermal) synthesis routes,
for the conversion of CH4 to desired products, have the
advantage that they circumvent the expensive and energy
intensive syngas step. They are currently, however,
technologically very challenging and costly, while only
achieving the same low yields.[4]
Nevertheless, the reforming of CH4 into syngas is
worldwide gaining increased attention, due to the
versatility of syngas for the production of many fuels and
chemicals, such as methanol, but also Fischer-Tropsch
fuels, H2, ethanol, dimethyl ether,… This results in a
growing interest for alternative (non-conventional)
reforming processes, like plasma technology. The
advantage of non-thermal plasmas is that the gas can
remain near room temperature while being “activated” by
electron impact excitation, ionization and dissociation
reactions. Both thermal and non-thermal plasma
reformers have already been developed.[6]
The DRM process, i.e.,
CH4(g) + CO2(g) 2 CO(g) + 2 H2(g))
is carried out at high temperatures by means of a catalyst.
On the industrial scale the DRM is carried out most
efficiently at 700°C, reaching thermodynamic equilibrium
conversions of CH4 and CO2 of 72% and 82%,
respectively. This results in an energy input of at least
about 3.42 eV per molecule and a corresponding
maximum theoretical energy efficiency of 58 %.[7]
In this paper, we study the plasma-based dry reforming
process in a dielectric barrier discharge (DBD) reactor, by
means of computer simulations and experiments. In our
investigation we make use of a zero-dimensional (0D)
chemical kinetics model. We carried out an extensive
study of the reaction chemistry while mimicking the
filamentary discharge regime. The model is applied to
long time-scales, corresponding to the typical residence
times of the gases in a DBD, in order to calculate the CH4
and CO2 conversion, the selectivity of the reaction
products and the energy efficiency of the process.
2. Description of the Model
2.1. 0D Chemical Kinetics Model
The computational model used in this work to describe
the plasma chemistry is a zero-dimensional (0D) kinetic
model, called Global_kin, developed by M. Kushner and
coworkers.[8] In this work the 0D plasma chemistry
module and the Boltzmann equation module are used. The
time-evolution of the species densities is calculated, based
on production and loss processes, as defined by the
chemical reactions. The rate coefficients of the heavy
particle reactions depend on gas temperature and are
calculated by Arrhenius equations. The rate coefficients
for the electron impact reactions are a function of the
electron temperature, and are calculated in the Boltzmann
equation module. Finally, the electron temperature is
calculated with an energy balance equation.
2.2. Plasma Chemistry Included in the Model
The plasma chemistry used in the model was
developed in previous work [7]. The model considers 57
different species, including the electrons, various
molecules, radicals, ions and excited species. All these
species are listed in Table 1. They react with each other in
498 reactions: 121 electron impact reactions, 87 ion
reactions and 290 neutral reactions.
Table 1. List of species included in the model for the
CH4/CO2 gas mixture.
Molecules
CH4, C2H6, C2H4, C2H2, C3H8, C3H6, C4H2,
H2, O2, CO2, CO, H2O, H2O2, CH2O,
CH3OH, CH3CHO, CH2CO
Charged
species
CH5+, CH4+, CH3+, CH2+, CH+, C2H6+,
C2H5+, C2H4+, C2H3+, C2H2+, O2+, O-,
H3O+, OH-, electrons
Radicals
CH3, CH2, CH, C, C2H5, C2H3, C2H, C3H7,
H, O, OH, HO2, CHO, CH2OH, CH3O,
C2HO, CH3CO, CH2CHO, C2H5O2
Excited
species
H2*, O*, CO2*, CO*, H2O*
2.3. Description of the DBD Setup in the Model
Since the model is zero-dimensional, we can only
simulate the plasma behavior as a function of time and we
cannot describe the spatial variation in a direct manner.
However, the temporal behavior can be translated into a
spatial behavior (as a function of distance along the DBD)
by means of the gas flow.
In the case of a CH4/CO2 plasma, a DBD typically
occurs in the so-called filamentary regime, consisting of a
large number of independent micro-discharge filaments.
In these micro-discharges a large fraction of the electron
energy is used for excitation, dissociation and ionization
of the molecules, and hence to initiate the chemical
reactions. This is the reason why including these microdischarges in the simulations is of prime importance for a
realistic description of the reaction chemistry. Again, we
cannot treat the spatial aspect of filament formation in our
0D model, but we can mimic the filamentary behavior by
simulating a large number of micro-discharge pulses as a
function of time. For more information about this
procedure, we refer to our previous work.[7], [9].
We considered triangular micro-discharge pulses of 30
ns, with a repetition frequency of 0.7 kHz. This
corresponds to an applied frequency of 35 kHz, as used in
the experiments, assuming that each molecule passes
through only one micro-discharge every 100 half cycles
(see detailed discussion in our previous work).[7] The
residence time for the experimental data is 6.8 s, as
calculated from the gas flow rate and the length of the
reactor. Finally, the maximum power deposition per pulse
is defined in such a way that the total specific energy
input (SEI) corresponds to the experimental values (i.e.,
in the order of 18-36 J∙cm-3; see below).
In experiments the SEI (typically expressed in kJ∙L-1 or
J∙cm-3) is defined as the applied power (W) divided by the
gas flow rate (L∙s-1). This value can be converted to the
input energy in eV per molecule, which gives us an idea
about the energy cost and energy efficiency of the process
under study:
Note that the value of 24.5
K and 1 atm.
is calculated for 298
Figure 1 illustrates the calculated electron density (Ne)
and electron temperature (T e) for one pulse as a function
of time. The calculated maximum E0/N is in the order of
200 Td. This results in a maximum Ne of ~3∙1013 cm–3 and
a maximum Te of ~3 eV during the pulse. At the start of
the pulse, Te reaches its maximum of ~3 eV, as the
electrons are heated by the electric field, whereas upon
pulse termination, Te drops significantly. Ne on the other
hand increases with time during the pulse and reaches its
maximum of 3×1013 cm-3, as the power leads to the
electron heating and subsequently it gives rise to electron
impact ionization, creating electrons during the pulse.
However, the Ne decays very slowly upon termination of
the pulse, indicating low recombination rates and/or the
fact that electrons might still be created in the early
afterglow by heavy particle reactions. For all investigated
operating conditions the maximum Te was around ~3 eV,
while the maximum Ne was in the order of 1012-1014,
which are typical conditions for a DBD.[10], [11]
Figure 1. Calculated electron density (black line, right
axis) and electron temperature (red line, left axis) during
one triangular discharge pulse of 30 ns for a 1:1 CH 4/CO2
mixture. The grey dashed lines indicate the start and the
end of the microdischarge pulse.
3.
Results
We apply the model to real time-scales, corresponding
to typical residence times of the gas molecules in the
plasma, in order to obtain conversions, selectivities and
energy efficiency, to be compared with experimental data.
It should be emphasized, however, that it is not yet the
focus of the present study to optimize the obtained
conversions, selectivities and energy efficiency. The latter
will be elaborated in future work.
afterglow, it contributes for 99% to the production of
CH4.
3.1. Conversion of CH4 and CO2:
The parameters of interest to define whether plasma
technology has enough perspectives for the dry reforming
of methane are, as already mentioned, the CO2 and CH4
conversion, the selectivity of the reaction products and the
energy efficiency of the process. The calculated
selectivity and energy efficiency will be presented in the
next sections. Here we will present the calculated
conversion, and compare with experimental data. The
conversion of CH4 and CO2 is defined as:
Figure 2. Calculated and experimental values of CH4 and
CO2 conversion as a function of specific energy input.
The calculated conversions of CH4 and CO2, are
plotted in figure 2 as a function of SEI, for a residence
time of 6.8 s. It is clear that the conversion of CH4 and
CO2 increases with the SEI, and the calculated conversion
of CH4 is about a factor of 2 higher than the conversion of
CO2. It is apparent from this figure that the conversion
increases more or less linearly with the SEI values, which
seems logical. However, a higher SEI value implies a
higher energy cost, and it has therefore a negative impact
on the energy efficiency (see section 3.3 below). There
will probably be an optimum between conversion,
selectivity and energy efficiency, which we will
investigate in future work.
During the pulse, the electron impact reactions are by
far the most important loss mechanisms for CH4, and
especially electron impact dissociation with the formation
of CH3 and H (i.e. e- + CH4  CH3 + H + e-) However,
immediately after pulse termination, the rates of these
electron impact reactions drop to nearly zero, due to the
drop in electron temperature (see figure 1 above). In the
afterglow, the chemical reactions with radicals and ions
are the most important loss processes. Integrated over one
pulse and afterglow, the electron impact reactions
contribute for 77 % to the loss of CH4, in spite of the fact
that they only occur during the nanoseconds pulse, while
the ion and radical reactions account for 23 %. The
formation processes of CH4 are not so relevant, because
CH4 is mainly lost (dissociated), but nevertheless, they are
briefly discussed here, as they have a negative
contribution to the conversion of CH4. During the pulse,
the three-body recombination reaction between CH3 and
H radicals, with a gas molecule as third body, is the only
production mechanism for CH4, and this reaction is also
dominant in the afterglow. Integrated over pulse and
It was observed that during the pulse the rate of the
electron impact reactions yielding loss of CH4 is one
order of magnitude higher than the rate of the major
production process, so that during the pulse mainly
dissociation of CH4 takes place. In the afterglow,
however, the formation processes are characterized by a
higher overall rate, so that the formation of CH4 is now
much larger than the loss, resulting in a rising CH4
density for the afterglow.
For CO2, the electron impact (dissociation and
ionization) processes are also dominant during the pulse.
Similar to CH4, the electron impact reaction rates drop to
zero in the afterglow, and the chemical reactions with
radicals become responsible for the loss of CO2.
Integrated over one pulse and afterglow, the electron
impact reactions contribute for 33% to the loss of CO2,
wheras the chemical reactions with radicals contribute for
67%. This is in contrast to CH4, where the electron impact
reactions were clearly dominant. The higher stability of
CO2 causes the conversion during the pulse to occur
predominantly by electron impact, while CH 4 loss is
initiated by both electrons and radicals, thus explaining
the higher conversion of CH4 versus CO2 during the
discharge pulse. In contrast to CH4, where the formation
mechanisms became more important than the loss
processes in the afterglow, for CO2 the chemical loss
processes are still far more important than the formation
processes. Hence, CO2 will continue to be lost in the
afterglow.
3.2. Selectivity of main products
The selectivity of the formed products is an even more
important quantity than the conversion, as we target the
formation of value-added chemicals or new fuels. H2 and
CO are formed with the higest selectivities, followed by
the sum of C3H8 and C3H6, CH2O, the sum of C2H6, C2H4
and C2H2 hydrocarbons and CH3OH. The selectivity of
the main products, H2 and CO, is defined as follows:
When comparing the calculated and experimental data
in Figure 3, it appears that the calculated H2 selectivity is
overestimated by about a factor 2, while the deviations in
CO selectivity are only minor, and very similar to the
deviation of the CO2 conversion. A more extensive
validation of the obtained selectivities will be carried out
in future work.
%, respectively, yielding an overall conversion of 12 %.
This gives rise to an energy efficiency of 7.3 %. Note that
this is still a factor of 8 lower than the classical dry
reforming process. However, the latter is not unexpected,
as the plasma-based dry reforming process is not yet
optimized. Moreover, it is well possible that a pure DBD
reactor will never be competitive with the classical dry
forming. Therefore, in future work, we will also
investigate the combination with catalysis, so-called
plasma catalysis, as well as the performance of other
types of plasma reactors, which are stated in literature to
be more energy efficient.[12]
Figure 4. Calculated (open symbols) and experimentally
measured (full black symbols) energy efficiency as a
function of specific energy input.
Figure 3. Calculated and experimental values of H2 and
CO selectivity as a function of specific energy input.
3.3. Energy efficiency
As mentioned above, the energy efficiency is probably the
most important criterion for the dry reforming process.
The thermodynamic energy cost for dry reforming is 2.56
eV per converted molecule. The energy cost for the
classical dry reforming process amounts to at least 3.42
eV/molecule, for a CH4 conversion of 72% and a CO2
conversion of 82%, hence corresponding to a maximum
achievable energy efficiency of 58%. Therefore, this
value should be the target of the plasma-based dry
reforming process. The energy efficiency ( ) is defined
here as follows:
where
is the conversion. In figure 4 the energy
efficiency is plotted as a function of SEI. It is clear that
the higher conversion with increasing SEI (see figure 2
above) does not compensate for the higher energy input
with regard to the energy efficiency. Indeed, the highest
energy efficiency is obtained for the lowest SEI value
considered in this work, i.e., 18 J/cm3. For this condition
the calculated CH4 and CO2 conversions were 16 % and 8
4. References
[1] Fourth Assessment Report: Climate Change 2007
Synthesis Report, Intergovernmental Panel on Climate
Change IPCC: Switzerland, Geneva, (2007);
[2] Sustainable Chemistry Strategic Research Agenda,
European Technology Platform for Sustainable
Chemistry: Belgium, Brussels, (2005);
[3] W. McDonough, M. Braungart, P. Anastas, J.
Zimmerman, Environmental Science and Technology,
37, 23, (2003).
[4] J. H. Lunsford, Catalysis Today, 63 (2000).
[5] J. R. H. Ross, Catalysis Today, 100, 1–2 (2005).
[6] G. Petitpas, et al., International Journal of Hydrogen
Energy, 32, 14 (2007).
[7] R. Snoeckx, R. Aerts, X. Tu, A. Bogaerts, The journal
of physical chemistry: C, 117, 10 (2013).
[8] R. Dorai, K. Hassouni, M. J. Kushner, Journal of
Applied Physics, 88, 10 (2000).
[9] R. Aerts, T. Martens, A. Bogaerts, The Journal of
Physical Chemistry C, 116, 44 (2012).
[10] A. Fridman, Plasma chemistry (2008).
[11] U. Kogelschatz, Plasma chemistry and plasma
processing, 23, 1 (2003).
[12] A. Gutsol, A. Rabinovich, and A. Fridman, Journal
of Physics D: Applied Physics, 44, 27 (2011).