Feliz 2021 J. Phys. D Appl. Phys. 54 334003

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Journal of Physics D: Applied Physics

PAPER You may also like


- Selective cytotoxic effect of non-thermal
Influence of ionic conductivity and dielectric micro-DBD plasma
Byung-Su Kwon, Eun Ha Choi, Boksoon
constant of the catalyst on DBD plasma-assisted Chang et al.

- Ultrasound generated by alternating


CO2 hydrogenation into methanol current dielectric barrier discharge plasma
in quiescent air
Xin Zhang, Y D Cui, Chien-Ming Jonathan
Tay et al.
To cite this article: Maxwell Quezada Feliz et al 2021 J. Phys. D: Appl. Phys. 54 334003
- An experimental study on the thermal
characteristics of NS-DBD plasma
actuation and application for aircraft icing
mitigation
Yang Liu, Cem Kolbakir, Andrey Y
View the article online for updates and enhancements. Starikovskiy et al.

This content was downloaded from IP address 163.239.188.159 on 14/11/2023 at 09:25


Journal of Physics D: Applied Physics

J. Phys. D: Appl. Phys. 54 (2021) 334003 (15pp) https://doi.org/10.1088/1361-6463/abfddd

Influence of ionic conductivity and


dielectric constant of the catalyst on
DBD plasma-assisted CO2
hydrogenation into methanol
Maxwell Quezada Feliz1,2, Isabelle Polaert2,∗, Alain Ledoux2, Christian Fernandez1
and Federico Azzolina-Jury1,∗
1
Normandie Univ, ENSICAEN, UNICAEN, CNRS, LCS, Laboratoire Catalyse et Spectrochimie, 14000
Caen, France
2
Normandie Univ, UNIROUEN, INSA Rouen, LSPC, Laboratoire de Sécurité des Procédés Chimiques,
76000 Rouen, France

E-mail: isabelle.polaert@insa-rouen.fr and federico.azzolina-jury@ensicaen.fr

Received 8 December 2020, revised 17 April 2021


Accepted for publication 4 May 2021
Published 4 June 2021

Abstract
Dielectric barrier discharge (DBD) plasma technology is a promising method for producing
methanol from CO2 hydrogenation as the reaction can be run at atmospheric pressure and
temperatures below 100 ◦ C. The choice of the catalyst is crucial and has to be made not only
according to its activity and selectivity towards the desired product, but its effect on plasma
properties. In this work, the influence of several important catalytic properties of DBD plasma
such as the dielectric constant of the catalyst and ionic conductivity is studied. The effects of the
catalyst support and the addition of promoters on CO2 hydrogenation under DBD plasma are
also studied. To this end, Cu and Cu–ZnO catalysts supported on γ-Al2 O3 and a template-free
seedless ZSM-5 (Si/Al molar ratio of 23) were prepared to study their catalytic performance on
CO2 hydrogenation into methanol under DBD plasma. These catalysts were fully characterized
by XRD, SEM, N2 physisorption, temperature programmed reduction and in situ FTIR CO
adsorption. The relative complex permittivity of the catalysts was measured and the ionic
conductivity was estimated using a modified Debye model. In this paper, the role of the ionic
conductivity of the catalyst was identified as a crucial parameter in plasma-assisted CO2
hydrogenation. It was found that the lower the value of the ionic conductivity, the better the CO2
conversion. Indeed, high ionic conductivity reduces the density of the plasma and decreases the
dissociation of CO2 . The highest CO2 conversion value (34.0%) was observed for the
nonconductive alumina support, whereas the highest methanol yield (0.5%) was observed for
the zeolite-supported Cu–ZnO catalyst.

Supplementary material for this article is available online


Keywords: DBD plasma, methanol synthesis, CO2 hydrogenation, dielectric constant,
ionic conductivity, ZSM-5

(Some figures may appear in colour only in the online journal)


Authors to whom any correspondence should be addressed.

1361-6463/21/334003+15$33.00 1 © 2021 IOP Publishing Ltd Printed in the UK


J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

1. Introduction Table 1. Possible reactions involved in methanol production from


carbon oxides.
Carbon dioxide is a major greenhouse gas, but it can also be ∆r H 0 298K ∆r G0 298K
seen as an abundant source of C1 . For decades, several propos- Reaction (kJ mol−1 ) (kJ mol−1 )
als to reduce anthropogenic CO2 emissions have been made
for recycling carbon dioxide into fuels. CO2 is a non toxic, CO2 + 3H2 ↔ CH3 OH + H2 O −49.51 3.30
CO2 + H2 ↔ CO + H2 O 41.17 28.64
non corrosive, non flammable [1] and very thermodynamic-
CO + 2H2 ↔ CH3 OH −90.68 −25.34
ally stable molecule (∆f G0 298K = −394 kJ mol−1 ), requiring a
large amount of energy for its dissociation (+283 kJ mol−1 or
2.9 eV molecule−1 ) [2]. CO2 is produced from almost all com-
bustion processes and is the main greenhouse gas (after water than the energy (or temperature) of the other species present
vapour). The concentration of CO2 in the atmosphere reached in the gas mixture. Electrons travelling to the positive elec-
417 ppm in March 2020, according to NOAA [3]. Currently, trode can interact with other plasma species or collide with
to prevent its release into the atmosphere, one of the existing gas molecules leading to ionization, vibrational excitation or
technologies is its capture and storage in underground reser- even dissociation of these molecules. The high density of these
voirs [4], but this technique does not benefit from CO2 as a car- highly energetic species allows plasma-assisted reactions to
bon source. Another approach, as proposed in this article, is the take place at relatively low temperatures. Dielectric barrier dis-
chemical utilization of this captured CO2 for its valorization charge (DBD) NTP reactors have been used to produce meth-
into hydrocarbons, in this particular case into methanol. The anol from CO2 hydrogenation at atmospheric pressure and
source of pure CO2 would be the exhaust gases from chem- temperatures below 100 ◦ C, which is fairly low when com-
ical industry that are not pure and which are purified via cap- pared to conventional heating methods [15–19]. However, sev-
ture technology already proposed in literature [5, 6]. The main eral key parameters related to plasma-assisted catalysis need
interest in making methanol is to create a source of chemical to be fully understood when aiming to optimize CO2 hydro-
energy, which would be used to store energy from renewable genation reactions.
sources [7]. Methanol is preferred to gaseous fuels (methane, In plasma-assisted catalysis, the plasma can modify the
hydrogen, etc) because, at ambient conditions, it is liquid and properties of the catalyst but the catalyst itself can also affect
therefore easily transportable, without the safety risks associ- the properties of the plasma. In a review by Neyts and Bogaerts
ated with the transportation of pressurized gases. Methanol is [20], several of these effects, which lead to a plasma-catalyst
also one of the most important products of the petrochemical synergy, were addressed. Among the effects of a catalyst on
and chemical industries, serving as raw material for the pro- the properties of the plasma, the following can be mentioned:
duction of formaldehyde, dimethyl ether [8], etc. the enhancement of the electric field, the modification of the
Methanol can be synthesized from carbon oxide hydrogen- discharge type and the formation of micro discharges within
ation reactions. Possible reactions involving methanol produc- the catalyst pores. Concerning the effects of the plasma on the
tion from carbon oxides are presented in table 1. Methanol can properties of the catalyst, we can cite the modification of the
be produced via CO2 hydrogenation (table 1, line 1). However, adsorption capacity of the catalyst, the surface area, oxidation
this reaction is accompanied by an undesired secondary endo- state of the metallic active phase, formation of hot spots and
thermic reaction (RWGS), which is favoured at high temperat- modification of the surface reaction pathway.
ures and consumes the same reactants (table 1, line 2). In both The synergetic effects observed in plasma catalysis can be
reactions, low CO2 conversion is usually achieved without tailored and plasma-assisted catalytic processes can be optim-
the use of a catalyst as a consequence of the high molecu- ized through a better understanding of the parameters govern-
lar stability of CO2 . Thus, both reactions are usually stud- ing the mutual interaction between the catalyst and the plasma
ied at high temperatures (>200 ◦ C) and pressures (>20 atm) [21]. In this regard, the effect of four magnitudes has mainly
over alumina-supported copper-based catalysts in convention- been studied: the dielectric constant of the catalyst, its ionic
ally heated fixed-bed reactors [1, 9–13]. Methanol can also be conductivity, its particle size and the porosity of the bed. Con-
obtained from CO hydrogenation (table 1, line 3), which is cerning the first one, Butterworth et al [22, 23] reported that an
the current reaction used in the industry. In this case, CO is increase in the dielectric constant of the catalyst decreases the
obtained either by methane reforming or through coal gasific- breakdown voltage. In their work, the authors described two
ation [9]. Carbon oxide hydrogenation can also produce meth- types of discharge modes: the polar mode and surface streamer
ane as a secondary product (refer to table 1 of Gao et al [14]). mode. For a given dielectric, the plasma discharge first occurs
A nonconventional approach to methanol synthesis from in polar mode and then switches to surface streamer mode
CO2 hydrogenation is the use of plasma technology (this upon voltage increase. If the dielectric constant is increased,
work). Plasma is the fourth state of matter, consisting of a mix- the necessary voltage for the transition between these two
ture of highly energetic species, such as electrons, ions, rad- modes is also increased. However, the authors also reported
icals and excited molecules. Among the numerous available that this transition can be related to the ionic conductivity and
technologies for the generation of plasma, nonthermal plasmas the porosity of the bed. The influence of the dielectric constant
(NTPs) are among the most used in plasma-assisted catalysis. has also been addressed by van Laer and Bogaerts [24]. Upon
NTPs are nonequilibrium plasmas since the electron energy simulation studies, the authors quoted that plasma relocates
(characterized by the electron temperature T e ) is much higher into the catalyst bed voids when the dielectric constant of the

2
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

material is increased. The authors claim that higher voltages Table 2. List of the catalysts used in this work.
are needed for plasma ignition when the material possesses a Metal content
low dielectric constant. They also addressed the influence of Si/Al molar
the particle size of the catalyst; the smaller the particle size, Support ratio Cu Zn Code
the higher the voltage required to ignite the plasma. Song et al γ-Al2 O3 — — — A
[25] found that the dielectric constant of the material influ- γ-Al2 O3 — 10% wt. — Cu/A
enced the plasma-catalyst synergetic effect on chlorobenzene γ-Al2 O3 — 10% wt. 10% wt. Cu–Zn/A
decomposition. The decomposition rate is higher in the case of ZSM-5 23 — — Z23
catalysts with high dielectric constant because the ionization ZSM-5 23 10% wt. — Cu/Z23
of the gas and the electron energy are enhanced by increas- ZSM-5 23 10% wt. 10% wt. Cu–Zn/Z23
ing the dielectric constant of the packed materials [26, 27].
Michielsen et al [28] studied CO2 dissociation in a packed-
bed DBD reactor using different dielectric materials and bead
sizes. The authors identified three important effects: (a) the 1310-73-2; NaAlO2 from Sigma Aldrich (Al2 O3 : 50%–56%
positive effect of the packing due to electric field enhancement, wt., Na2 O: 37%–45% wt.) CAS No. 11138-49-1 as the alu-
(b) the negative effect of the packing due to the reduction of minium source and distilled water. The Si/Al molar ratio of
the reactant residence time and (c) the influence of the packed- the prepared ZSM-5 was 23.
bed porosity with a positive or negative effect depending on The γ-alumina supports were obtained from commer-
the dielectric constant of the material inside the DBD reactor. cially available boehmite DISPAL 23N4-80 (Sasol) with large
Finally, Jo et al [29] studied the effect of the catalyst conduct- particles (50 mm). Boehmite was dehydrated through calcina-
ivity on methane production using an alumina-supported plat- tion at 550 ◦ C for 3 h, as reported elsewhere [36].
inum catalyst. They found that the presence of Pt decreases The prepared supports were then shaped into cylindrical
the magnitude of the electric field and thus, the methane yield. extrudates (5 mm in length and 2 mm in width) using boehmite
Based on these results, they inferred that the conductivity of as a binder [36]. The binder used in the manufacture of the
the catalyst was responsible for the decrease in magnitude of extrudates (boehmite) was transformed into γ-alumina during
the electric field. They also suggested that further studies using the calcination step. Therefore, zeolite extrudates also con-
conductive and nonconductive solids were needed to better tained alumina. In the case of the alumina support, the extrud-
identify the aforementioned effects. ates were made of pure γ-alumina only. The reason for the
From the previous paragraph, it is clear that even though preparation of these extrudates is that they were also used in a
several parameters have been observed to be influential conventionally heated fixed-bed setup to study the kinetics of
(dielectric constant, ionic conductivity, pellet size and bed CO2 hydrogenation to methanol under pressure in a pilot plant
porosity), their effects are yet to be fully understood. In this [37, 38]. For the catalytic tests presented in this work, a final
regard, we have chosen to study the effects of the ionic catalyst particle size between 400–800 µm was used and was
conductivity and dielectric constant on CO2 hydrogenation obtained by grinding the aforementioned extrudates.
into methanol in DBD plasma using two types of supports: Copper (10% wt.) and zinc (10% wt.) were added to the
γ-alumina and a ZSM-5 with Si/Al molar ratio of 23. The γ- alumina and zeolite extrudates by wet impregnation. Cop-
alumina support is the traditional support for this reaction due per nitrate trihydrate, Cu(NO3 )2 ·3H2 O from Sigma-Aldrich
to its amphoteric properties [30] while zeolites are good poten- (⩾99%), CAS No. 10031-43-3 and zinc nitrate hexahydrate,
tial candidates because of their high surface area and catalytic Zn(NO3 )2 ·6H2 O were used as metal precursors. In the case
activity [31–33]. These two supports were also chosen because of the bimetallic catalysts, both metal precursors were added
they differ in their ionic conductivity and dielectric properties simultaneously. Liquid removal was performed under vacuum
[34, 35]. These supports were impregnated with copper (Cu) at 80 ◦ C overnight and the dried catalysts were calcined under
and copper-zinc oxide (Cu–ZnO) to produce four catalysts and air at 600 ◦ C for 3 h. The list of catalysts prepared for this work
were fully characterized and tested in DBD plasma to ascertain is summarized in table 2.
the influence of the two aforementioned parameters. Finally,
the role of copper and zinc oxide active sites in CO2 hydro-
genation reactions is also addressed.
2.2. Catalyst characterization

The alumina and ZSM-5 structures (powder and extrudates)


2. Experimental
were verified using a PANalytical X’Pert PRO diffractometer
with CuKα radiation (λ = 0.154 18 nm, 40 mA, 45 kV).
2.1. Catalyst preparation
The XRD patterns were recorded between 2θ = 5◦ − 75◦
The MFI zeolite support used in this work was prepared in for the alumina catalysts and 2θ = 5◦ − 60◦ for the MFI
its sodium form (Na-ZSM-5) following a template-free seed- catalysts, with a step size of 0.016 73◦ and a time per step
less method [8]. The following reactants were used: LUDOX of 40 s.
HS-30 Colloidal Silica from Sigma Aldrich (30% wt. SiO2 The size and morphology of the catalysts were examined
suspension in water), CAS No. 7631-86-9, as the silica source; using a Philips XL-30 SEM with an acceleration voltage
NaOH pellets from VWR Chemicals (⩾98% assay), CAS No. of 30 kV. The results of these first two characterization

3
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Figure 1. Plasma DBD reactor (left) and conventional packed-bed reactor (right) for CO2 hydrogenation to methanol.

tests (XRD and SEM) allowed us to identify and valid- 5 ◦ C min−1 . Once the desired temperature was reached, small
ate the adequate crystallinity and morphology of the pre- volume doses (1.57 cm3 ) of CO were sent to the catalyst wafers
pared catalysts, as well as the presence of the desired metals. until complete surface saturation. The CO was used to study
Their corresponding figures are available in the supplementary the dispersion of the metallic active phase and to identify the
information provided (available online at stacks.iop.org/JPD/ different copper species (Cu0 , Cu+ , Cu2+ ). The CO adsorp-
54/334003/mmedia) (figures 1 and 2). tion was performed at room temperature.
The metal content of the synthesized catalysts was determ- The complex permittivity of catalysts can be measured by
ined by inductively coupled plasma (ICP). two techniques: the method of small perturbations in the res-
Nitrogen adsorption-desorption isotherms were measured onant cavity or the reflection method. The former is gener-
at 77 K with a Micrometrics Model ASAP 2020 volumetric ally used to measure permittivities at a given frequency as
adsorption analyzer. The corresponding figures are available a function of temperature and has been developed specific-
in the supplementary information. Before each analysis, the ally for microwave frequencies [39]. The latter is frequency-
catalysts were outgassed at 220 ◦ C overnight. The specific sur- dependent and allows the precise characterization of solids
face area of the catalysts was determined from the Brunauer– in a wider frequency domain. From these measurements, the
Emmett–Teller (BET) equation. Their total pore volume was static and infinite dielectric constants, as well as the ionic con-
estimated from the N2 adsorbed volume at P/P0 = 0.99. ductivity, can be calculated using an appropriate mathematical
The reducibility of the catalysts was determined through model. This allows us to evidence physical phenomena occur-
temperature-programmed reduction (TPR) analysis in a ring at the microscopic scale, such as dielectric relaxation.
Micromeritics AutoChem II analyzer. Samples of 50 mg of In this work, the complex permittivities of the prepared cata-
catalysts were loaded into a quartz U-tube. TPR was per- lysts were measured by a reflection technique described else-
formed using 5% v/v H2 in Ar with a total flow rate of where [40, 41]. Complex permittivity data were acquired in the
40 ml min−1 at standard temperature and pressure (STP) from microwave range between 0.5–20 GHz using a vector network
room temperature up to 1000 ◦ C with a heating ramp of analyzer from Agilent 2-PortPNA-L of the 5230 A series con-
10 ◦ C min-1 . nected to a high-temperature open-ended coaxial probe. The
In situ FTIR measurements were carried out using carbon measurements were carried out at room temperature (20 ◦ C).
monoxide as a probe molecule with a NicoletMagna 550 FTIR Prior to the measurements, the catalysts were sieved in order to
spectrometer equipped with an MCT detector. The FTIR spec- have a particle size of less than 100 µm. For the measurement
tra were recorded from 4000 to 600 cm−1 with an optical resol- of their dielectric properties, the catalysts were manually com-
ution of 4 cm−1 and 32 scans. Then, the catalysts were pressed pacted in a cylindrical polytetrafluoroethylene block of 20 mm
(2 × 108 Pa) into self-supported wafers (2 cm2 , ∼20 mg). Prior (height) × 20 mm (width). The complex permittivity measure-
to all infrared (IR) measurements, the catalysts were first activ- ments were repeated three times for each catalyst.
ated under secondary vacuum (6.10−4 Pa) at 350 ◦ C for 4 h
with a heating rate of 5 ◦ C min−1 . After activation, the tem- 2.3. Catalytic tests
perature was raised to 400 ◦ C using the same heating rate.
Then, the catalysts were consecutively reduced twice (with The experimental set-up used in this work to evaluate the per-
an in between evacuation of the formed water) for 30 min formance of the prepared catalysts on CO2 hydrogenation to
under 100 Torr of H2 . After the reduction step, the temper- methanol under both DBD plasma and conventional heating is
ature was reduced to room temperature using a cooling rate of presented in figure 1.

4
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Both conventional and DBD plasma reactors were fed with


40 ml min−1 (STP) of CO2 (3.75% v/v), H2 (11.25% v/v)
and Ar (85% v/v) using Brooks mass flow controllers at atmo-
spheric pressure. A CO2 :H2 stoichiometric molar ratio of three
was adopted for this study. Both reactors were made of quartz
and had the same dimensions: an inner diameter, outer dia-
meter and length of 1.0, 1.3 and 13 cm, respectively. Reaction
products were identified and quantified using a Bruker vertex
80v FTIR spectrometer with an optical resolution of 0.2 cm−1
and 16 scans, between 4000–600 cm−1 and a quadrupole mass
spectrometer (Pfeiffer Omnistar GSD 301). The gases at the
reactor outlet were heated at 120 ◦ C to avoid methanol con-
densation before analysis.
All the catalyst extrudates prepared in this work were
ground to a size between 400–800 µm for all catalytic tests. Figure 2. CO2 conversion versus temperature under conventional
A mass of 2 g of catalyst was used for both plasma-assisted heating.
and conventional heating tests. The height of the catalytic bed
was approximately 3 cm for all cases. Prior to each catalytic
test, the catalysts were reduced at 400 ◦ C with an H2 /N2 mix- where ṅ0 represents the inlet flow rate of a given spe-
ture (5% v/v) for 3 h. cies and ṅ the flow rate of that species measured at the
In the case of plasma-assisted reactions, a 0.5 cm dia- reactor outlet. The molar flow of the different species is
meter high-voltage electrode made of iron (Fe) was used. calculated according to the gas chromatography results. For
A 9 cm long hollow cylinder (copper), wrapping the quartz each compound, a specific calibration is done just before
reactor, , was used as a ground electrode. Catalyst particles the tests using gas calibration standards. The experimental
were held in place by a porous glass frit and placed within error was estimated at 5% by repeating the experiments three
the plasma discharge zone. The gap between the electrodes times.
was equal to 2.5 mm. The plasma discharge was generated
at atmospheric pressure, using a plasma function generator
3. Results
(FI5350GA) coupled to a high-voltage amplifier (Trek model
20/20 C). A sine voltage wave of 20 kV (peak-to-peak) at a fre-
3.1. Catalytic tests
quency of 9 kHz was adopted for all plasma-assisted catalytic
tests. These operating conditions were found to be optimal dur- 3.1.1. Conventional heating catalytic tests. Prior to the tests
ing a preliminary testing phase performed with the same DBD under DBD plasma, a few experiments were performed at the
reactor. Finally, the temperature inside the reaction zone was same pressure and gas flow but under conventional heating to
measured by manually inserting a thermocouple right after verify whether methanol was produced. These tests were car-
plasma extinction. ried out without a catalyst (empty reactor), with the alumina
In the case of the tests under conventional heating, the support and with the Cu/A. They were performed at atmo-
reactor was placed inside a furnace to regulate the reaction spheric pressure between 20 ◦ C–350 ◦ C. The CO2 conversion
temperature. A thermocouple was placed inside the catalytic as a function of the reaction temperature is shown in figure 2
bed to measure the temperature at its centre during the reac- for the three studied cases. The relative experimental error of
tion. Conventional heating catalytic tests were performed at these measures (and of all future measures presented in this
seven different temperatures: 25 ◦ C, 75 ◦ C, 150 ◦ C, 200 ◦ C, work) was 5%.
250 ◦ C, 300 ◦ C and 350 ◦ C. A heating ramp of 3 ◦ C min−1 In all three cases, CO2 conversion increased with tem-
was adopted to adjust the reaction temperature to the desired perature. The reaction products were CO and H2 O only and
value. they were produced through the endothermic reverse water
The CO2 conversion and the corresponding product gas shift (RWGS) reaction (table 1, line 2). Compared to the
selectivity and yields were calculated as follows: empty reactor, CO2 was converted into CO at lower tem-
peratures when the alumina catalysts (A and Cu/A) were
ṅ0CO2 − ṅCO2 used. The alumina support was able to catalyze the RWGS
XCO2 (%) = × 100, (1) reaction via its basic sites (oxygen of alumina structure).
ṅ0CO2
Both CO2 conversion and CO yield were enhanced after
ṅproduct Cu addition into the alumina support, evidencing the role
Sproduct (%) = × 100, (2) of copper as an active metal for this reaction. In all cases,
ṅ0CO2 − ṅCO2
no methanol was produced under conventional heating at
atmospheric pressure. Based on these results, no other cata-
ṅproduct lytic tests under conventional heating were pursued in this
Yproduct [%] = ∗ 100, (3)
ṅ0CO2 work.

5
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Figure 3. Results of the catalytic tests in DBD plasma for the alumina and Z23 supports (ER: empty reactor).

Table 3. Textural properties of alumina and zeolite supports and catalysts.

Microporous Mesoporous
surface area surface area Total surface Total pore volume Average pore Al2 O3 content (%
Catalyst (m2 g−1 ) (m2 g−1 ) area (m2 g−1 ) (cm3 g−1 ) size (nm) wt.) in extrudates

A 0 190 190 0.43 12.3


Cu/A 0 164 164 0.38 10.4 100
Cu–Zn/A 0 143 143 0.32 9.7
Z23 146 99 245 0.27 7.3
Cu/Z23 128 64 192 0.22 6.9 17
Cu–Zn/Z23 125 60 185 0.20 6.1

3.1.2. Plasma DBD catalytic tests. Figure 3 shows the to methane selectivity. On the other hand, CO2 conversion
CO2 conversion and the selectivity towards methanol and remained the same after the introduction of the Z23 support
methane obtained during the plasma-assisted catalytic tests. (when compared to the empty reactor), whilethe selectivity
As a reference, the results obtained under the same exper- of both products decreased. A slight increase in methanol
imental conditions but without any solid (with an empty selectivity is observed after copper addition, accompanied by
reactor) are shown. CO, methanol and methane were the a decrease in CO2 conversion. Finally, CO2 conversion and
only carbon-based products detected and were quantified the selectivity of both products increased after ZnO addition
at the outlet of the plasma reactor. The CO selectivity is for the CuZn/Z23.
not displayed, but it can be deducted from the other two In order to better understand and explain the observed beha-
as no other carbon products were detected. Thus, the CO viour, the following (sections 3.2–3.4) will provide the results
selectivities were: A (95.2%) > Z23 (94.9%) > Cu/Z23 of the performed characterization tests. Finally, section 4 of
(94.6%) > Cu/A (93.9%) > CuZn/A (93.6%) > empty reactor this work is devoted to the discussion of these characteriza-
(92.9%) > CuZn/Z23 (92.6%). The temperature of the sys- tion results and their relationship to figure 3.
tem was measured right after plasma extinction by inserting
a thermocouple into the centre of the catalyst bed. Its value
3.2. Physicochemical characterization of the catalysts
fluctuated between 83 ◦ C and 87 ◦ C, serving as an indic-
ator of the average temperature of the catalyst bed during the 3.2.1. BET surface area. The textural properties of alumina
reaction. and zeolite supports and catalysts are presented in table 3. The
First of all, it can be observed that CO2 is converted into alumina content of alumina and zeolite catalyst extrudates was
methanol, methane and carbon monoxide in the absence of estimated by ICP and is also available. For all supports, the
a support/catalyst (empty reactor). After the introduction of introduction of either copper or copper and zinc reduced the
alumina, CO2 conversion increased at the cost of a decrease total surface area. In addition, the total pore volume and aver-
in both methanol and methane production. Copper addition to age pore size of all supports were found to be diminished after
the alumina support decreased CO2 conversion while increas- the introduction of the metal(s), indicating a partial blocking of
ing both methanol and methane selectivity. Similar behaviour the pores of the support. The Z23 presented the highest surface
is observed after ZnO addition (CuZn/A), but only with respect area whether as a standalone support or as a catalyst. Better

6
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Table 4. Catalyst metal content and hydrogen uptake. in figures 4(b) and (c). The high-temperature reduction peaks,
H2 /Cu CuO/Cu2 O located at 513.5 ◦ C and 629.2 ◦ C for the alumina and Z23
Catalyst % wt. Cu % wt. Zn mole ratio mole ratio catalysts, respectively, are ascribed to the reduction of mixed
CuAl2 O4 and ZnAl2 O4 spinels [42]. This can be explained by
Cu/A 8.5 — 0.71 0.5 the introduction of a higher total amount of metal into the sup-
Cu–Zn/A 7.8 7.1 0.76 6.7 ports, which allowed the formation of spinels. On the other
Cu/Z23 9.8 — 0.98 0.3 hand, the pre-reduction CuO/Cu2 O mole ratio was consider-
Cu–Zn/Z23 10.0 10.0 0.80 4.9
ably modified after the addition of ZnO, indicating the pres-
ence of the more easily reducible Cu2+ species. This last fact
suggests that ZnO addition lowered the interaction between
dispersion of the active phase may have been achieved in the
the CuO particles and the supports.
case of the Z23 catalysts.

3.2.2. Temperature programmed reduction (TPR). The TPR 3.2.3. CO adsorption. Carbon monoxide was used as a
profiles of all catalysts are shown in figure 3. The H2 /Cu mole probe molecule to study the copper species in the alumina
ratio (H2 uptake) and catalyst metal content, determined by and zeolite catalysts. The CO IR bands were normalized with
ICP, are presented in table 4. respect to the total amount of CO adsorbed at full surface
First, the effect of the support can be analyzed through coverage. The complete surface coverage for the Cu/A and
comparison of the monometallic copper catalysts (Cu/A and Cu/Z23 was reached at 170 and 210 µmol CO/g, respectively.
Cu/Z23) in figure 4(a). For both supports, two reduction peaks The unprocessed CO adsorption IR spectra are shown in the
were identified by deconvolution. In the case of alumina sup- left side of figure 5, while the corresponding smoothed and
ports, CuO reduction presented two peaks located at 230.0 ◦ C deconvoluted spectra are shown to the right. In the case of the
and 270.1 ◦ C. These reduction peaks can be ascribed to the Z23, deconvolution was stopped at 2090 cm−1 because the IR
stepwise reduction of CuO to Cu2 O (Cu2+ to Cu+ ) followed bands after this value corresponded to the Z23 support itself
by the reduction of Cu2 O to metallic copper (Cu+ to Cu0 ), and thus did not vary with CO dosing.
respectively [42, 43]. In the case of the zeolite support, a con- Several IR bands were observed in the case of the alumina
voluted CuO reduction peak at 307 ◦ C was observed. This catalyst. The bands at 2071 and 2097 cm−1 are assigned to car-
peak is the contribution of the two CuO reduction peaks loc- bonyls on metallic copper (CO-Cu0 species) [46–48]. The IR
ated at 268.7 ◦ C and 307.3 ◦ C, respectively. It can be observed bands at 2140 and 2120 cm−1 are ascribed to CO adsorbed on
that, overall, lower temperatures are needed for CuO reduction Cu+ species [46–48]. The difference between these two bands
into metallic copper (Cu2+ to Cu0 ) in the case of the alumina is based on the oxygen coordination of Cu+ species: tetragonal
support. This can be explained by the interaction between the at 2120 cm−1 associated with ordinary Cu2 O phase and tetra-
CuO and the support. This interaction can be observed by ana- gonal pyramid at 2140 cm−1 as a consequence of the interac-
lyzing the pre-reduction CuO/Cu2 O mole ratio of the catalysts tion of CuO and Cu2 O phases with support oxygen ions [46].
(table 4), which was estimated by dividing the area of the cor- The IR bands above 2140 cm−1 are usually assigned to linear
responding CuO and Cu2 O peaks during reduction in figure 4. carbonyls over Cu2+ species [48]. It is worth mentioning that
This ratio was of 0.5 and 0.3 for alumina and zeolite supports, no IR bands in the range of 2200–2245 cm−1 were observed,
respectively. More easily reducible CuO species (Cu2+ ) can indicating the absence of CO adsorption over alumina support
be obtained with the alumina support, which can be an indic- through dative ligation to uncoordinated Al3 + cations [47].
ator of lower interaction between the introduced CuO and the In the case of zeolite catalysts, only the CO adsorption IR
alumina support. It is worth mentioning that for the mono- bands associated with Cu+ and Cu0 species were observed.
metallic copper catalysts (figure 4(a)), H2 consumption was The IR bands at 2104–2019 cm−1 and 2121–2126 cm−1 can
stopped after 380 ◦ C. This indicates the absence of copper be ascribed to CO adsorption on reduced copper (CO-Cu0 )
spinels (CuAl2 O4 ), which are reduced in the temperature range [49, 50]. The well-known IR band at 2158 cm−1 is assigned
of 550 ◦ C–800 ◦ C [42, 44]. to Cu+ -CO species (mono or bicarbonyls). The initial CuO
Finally, the effect of adding ZnO to the Cu catalysts is (Cu2+ ) species observed at the beginning of the TPR profiles
addressed. Copper reducibility was increased for the alumina did not appear after in situ catalyst reduction in the case of Z23,
catalyst, where H2 uptake was increased by 0.05, in accord- indicating that all Cu+2 species were either partially (Cu+ ) or
ance with the available literature [45]. Nevertheless, H2 uptake totally (Cu0 ) reduced. This can be explained by the higher H2
decreased by 0.18 in the case of the Z23 catalyst. This could be uptake (and thus copper reducibility) of the Cu/Z23 catalyst.
explained by the higher metal (Cu and Zn) content of this cata- However, in the case of alumina catalyst, all three Cu2+ , Cu+
lyst when compared to the alumina one (table 4). This means and Cu0 species were observed.
that ZnO can enhance Cu reducibility only if the total metal The influence of ZnO addition on Cu catalysts could not
content remains lower than approximately 15% wt. When ZnO be studied because of the low IR absorbance intensity of the
was added, the temperature and area of the two CuO reduc- Cu–ZnO samples caused by the relatively high metal loading.
tion peaks were modified. For both supports, the two CuO Copper dispersion and crystallite size for the two cata-
reduction peaks were shifted towards higher temperatures. lysts could be estimated through the integration of the IR
Moreover, the presence of a new reduction peak is observed bands associated with the CO-Cu0 complexes. Quantification

7
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Figure 4. TPR profiles of the catalysts.

Figure 5. CO adsorption over the Cu/A and Cu/Z23.

8
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Table 5. Characterization of metallic copper over alumina and increased slightly (+6%), while the dielectric loss factor did
zeolite supports. not experience any significant variation. These two different
Catalyst Cu/A Cu/Z23 kinds of behaviour may highlight important differences in the
dielectric properties of these two materials. Next, several con-
Cu dispersion (%) 9.4 13.7 (min.) clusions can be drawn from the Argand diagram of the Z23
Cu crystallite average size (nm) 10.5 7.2 catalysts (figure 6(c)). First, the initial shape of the curve of
all three solids can be approximated to a semicircular arc,
indicating the presence of relaxation processes. Second, after
of Cu2+ , Cu+ and Cu0 species was performed using the molar this semicircular portion, the dielectric loss factor increases
extinction coefficients proposed by Kohler et al for these three as the value of the dielectric constant increases, which is a
species: εCu2+ = 1.16 cm mol−1 , εCu+ = 0.72 cm mol−1 and clear indicator of ionic conductivity [40]. In zeolites, such
εCu0 = 0.39 cm mol−1 [51]. The details of the calculations are as the one synthesized in this work, this conductivity can be
provided in the supplementary information. For the Cu crys- attributed to the Na+ ions present in the zeolite framework.
tallite size estimation, a crystallite cubic shape was adopted Finally, the different overall values obtained for the curves
and a Cu density of 8.96 g cm−3 was used. The copper disper- shown indicate different dielectric behaviour for Z23, Cu/Z23
sion and crystallite size of the two catalysts are presented in and CuZn/Z23. Concerning the alumina (figure 6(d)), it shares
table 5. the first and third conclusion with the Z23 catalysts, but not the
The highest copper dispersion as well as the smallest Cu second. After the corresponding semicircular portion, it can be
particle size were both obtained with the Cu/Z23. These prop- observed that, as the value of the dielectric constant increases,
erties are closely related to the support itself. The alumina the dielectric loss factor only decreases. This is a clear indic-
support favoured the creation of more easily reducible copper ator of the absence of ionic conductivity and is present for all
species (Cu2+ ), but it also presented low copper reducibility the alumina catalysts, representing a key distinction between
and poor dispersion due to the relatively large metallic copper them and the Z23 catalysts.
particles present after catalyst reduction. To better understand the implications of the aforementioned
The results from this section can be summarized as follows. differences between the catalysts and how they would affect
The specific surface area (with and without metal) was higher CO2 hydrogenation, their quantification is necessary. In this
for Z23, while the alumina catalysts presented higher pore regard, a Debye model has been used and is presented in the
volumes. The TPR analysis showed that the addition of ZnO to following section.
the copper catalysts produced two effects: (a) enhancement of
Cu reducibility as long as the metal (Cu and ZnO) loading was
3.4. Estimation of the dielectric constant and ionic
lower than 15% wt. and (b) modification of the pre-reduction conductivity of the catalysts
CuO/Cu2 O molar ratio. The CO adsorption over the copper
catalysts allowed the estimation of Cu dispersion and particle For the quantification of the desired parameters, the real and
size. Cu dispersion was found to be higher for Cu/Z23 than imaginary parts of the relative complex permittivity were fit-
Cu/A, and Cu/Z23 presented a smaller crystallite size. ted using a modified Debye model. This model includes two
relaxation times as well as a term for the ionic conductivity,
as is often recommended when working with zeolitic supports
3.3. Measurement of the relative complex permittivity of the [40, 52]. The equations were as follows:
catalysts
εS1 − ε∞ εS2 − ε∞
The relative complex permittivity, εbr (T, ω) = εr ′ (T, ω) + εr ′ = ε∞ + + , (5)
1 + ω 2 τ12 1 + ω 2 τ22
iεr ′ ′ (T, ω), of a material is composed of two parts. Its real part
(εr ′ ) is related to the capacity of the material to store energy σ (εS1 − ε∞ ) ωτ1 (εS2 − ε∞ ) ωτ2
when submitted to an external electric field. On the other hand, εr ′ ′ = + + , (6)
ωε0 1 + ω 2 τ12 1 + ω 2 τ22
its imaginary part (εr ′ ′ ) is related to the capacity of the mater-
ial to either convert into heat or dissipate the received energy where ω (rad s−1 ) is the wave pulsation, ε∞ is the permittivity
when submitted to an external electric field. Figure 6 shows the at an infinite frequency, εS is the static permittivity, τ (1/2πf )
measurements of both parts of the relative complex permittiv- is the relaxation time (s) and σ is the conductivity (s m−1 ).
ity of the prepared alumina and Z23 catalysts as a function The imaginary part of the complex permittivity of all prepared
of frequency between 0.5–20 GHz at 20 ◦ C. By computing supports and catalysts is shown in figure 6. The real part of
repeatability experiments, the maximum relative errors on the their respective complex permittivities is presented in supple-
reported dielectric properties were ∆εr ′ = 5% and ∆εr ′ ′ = 8%. mentary figure 4.
Figure 6(a) shows that the dielectric constant of Z23 In figure 7, several contribution terms were observed upon
increased slightly after the addition of both Cu and ZnO, while deconvolution of the curves of the imaginary part of the com-
the dielectric loss factor remained practically unchanged. In plex permittivity: (a) two main relaxation phenomena for both
figure 6(b), different behaviour was observed for the alumina Z23 and alumina catalysts and (b) a conductivity term, but
catalysts. After copper addition, both the dielectric constant only for the Z23 catalysts. In the microwave frequency range,
and the dielectric loss factor decreased (11% and36%, respect- both catalysts presented a high relaxation frequency f 1 loc-
ively). After ZnO addition, the dielectric constant of Cu/A ated at approximately 12 GHz, which can be ascribed to the

9
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Figure 6. (a) Dielectric constant and dielectric loss factor of the Z23 catalysts as a function of frequency, (b) dielectric constant and
dielectric loss factor of the alumina catalysts as a function of frequency, (c) Argand diagram of the Z23 catalysts and (d) Argand diagram of
the alumina catalysts.

rotational polarization of adsorbed water molecules. A second- is significant for Z23. It was improved after Cu addition and
ary low relaxation frequency f 2 was also observed at approx- then decreased upon ZnO introduction, the latter probably due
imately 1.5 GHz probably due to an interfacial polarization to the steric effect caused by the increased metal loading.
mechanism. Overall, the measurements of the complex permittivity
The obtained fitted values for all supports and catalysts provided very important information about the dielectric prop-
are presented in table 6. The static dielectric constants calcu- erties of the prepared catalysts. In this regard, the two static
lated in this study are particularly relevant because they are dielectric constants (εS1 and εS2 ) associated with the two main
the dielectric constant values of the catalysts when used at low relaxation phenomena of the Z23 catalysts were found to be
frequencies (such as 9 kHz), which are those of DBD plasma. slightly higher than those of the alumina catalysts. This indic-
The conductivity thus determined can be assimilated to dir- ates that the Z23 catalysts have a slightly higher capacity to
ect current conductivity and does not depend on the frequency store energy than their alumina counterparts. On the other
for many materials and in particular ionic materials [53]. The hand, ionic conductivity (at low frequency) was observed in
values of the direct current conductivity, obtained by meas- the case of the zeolite catalysts only. Consequently, the alu-
uring dielectric losses over a frequency interval around GHz, mina catalysts should present a better dielectric character than
are therefore applicable to the characteristic frequencies of zeolites in the DBD plasma frequency range.
the plasma used here. Thus, the identification of phenomena
occurring at low frequency is key as they will be predominant
under the conditions used in this work. 4. Discussion
In table 6, both the static and infinite dielectric constants εS
and ε∞ of the alumina catalysts first decreased after the addi- The effect of the properties analyzed of sections 3.2–3.4
tion of Cu and then increased after ZnO addition. However, over CO2 conversion and product selectivity is discussed and
the same trend could not be confirmed in the case of the Z23 presented below. Table 7 summarizes the CO2 conversion
catalysts, most likely due to the different metal/support inter- andproduct yields of the catalysts discussed in this work.
action discussed in section 3.2.2. Overall, the zeolite catalysts
presented higher values of εS and thus a slightly higher capa- 4.1. CO2 conversion
city to store energy than their alumina counterparts.
Following the experimental trend (figures 6(c) and (d)), the Figure 3 evidences that carbon monoxide was the major spe-
term corresponding to the ionic conductivity was not evalu- cies issued from CO2 conversion. It was shown in section 3.1.1
ated for the alumina catalysts. In contrast, ionic conductivity that CO was produced through the RWGS reaction over a

10
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Figure 7. Imaginary part of the relative complex permittivity of the alumina and Z23 catalysts versus frequency.

Table 6. Complex permittivity terms obtained from the modified Debye model for alumina and Z23 catalysts.

Model parameter
Catalyst ε∞ εS1 εS2 f 1 (GHz) f 2 (GHz) σ (s m−1 )

A 2.37 3.09 2.83 11.1 1.6 0


Cu/A 2.26 2.77 2.57 12.3 1.5 0
Cu–Zn/A 2.40 2.98 2.70 12.2 1.0 0
Z23 2.46 3.13 2.79 12.3 1.4 7.1.10−3
Cu/Z23 2.65 3.25 3.03 12.6 1.2 9.9.10−3
Cu–Zn/Z23 2.69 3.39 3.05 12.2 1.2 7.8.10−3

catalyst, but only at temperatures higher than those reached no surprise that the physico-chemical properties of the cata-
under plasma assistance. This means that most of the CO is lysts themselves are not able to provide enough information
not formed over the catalyst but in the gas phase. According to explain the observed CO2 conversion trends. For instance,
to the work of Wang et al [18], this occurs mainly by CO2 the Z23 support presented a higher surface area than the alu-
electron impact dissociation, with a minor contribution from mina one (table 3), yet a lower CO2 conversion (figure 3). After
electron impact vibrational excitation. With this in mind, it is copper addition, the surface area of both supports decreased,

11
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

Table 7. CO2 conversions and product yields of the supports and and increased, respectively. It would then seem that CO2 con-
catalysts. version is somewhat favoured by the presence of non reduced
Catalyst X CO2 (%) Y MeOH (%) Y CH4 (%) Y CO (%) copper species. As both copper species are active sites for CO2
hydrogenation into methanol [12, 57–60], it is difficult to pre-
Empty reactor 25 ± 1.2 0.3 ± 0.01 1.5 ± 0.07 23.4 ± 1.2 cisely establish why unreduced copper species would better
A 34 ± 1.7 0.2 ± 0.01 1.4 ± 0.07 32.7 ± 1.6 catalyze CO2 under our experimental conditions. Both reduced
Cu/A 28 ± 1.4 0.4 ± 0.02 1.4 ± 0.07 26.6 ± 1.3
and unreduced species were evidenced by CO adsorption tests.
CuZn/A 27 ± 1.3 0.4 ± 0.02 1.4 ± 0.07 25.4 ± 1.3
Z23 25 ± 1.2 0.1 ± 0.00 1.1 ± 0.05 23.8 ± 1.2
These tests also allowed us to estimate the copper dispersion,
Cu/Z23 21 ± 1.0 0.2 ± 0.01 1.0 ± 0.05 19.7 ± 1.0 where the Z23 showed a higher value. However, this higher
CuZn/Z23 24 ± 1.2 0.5 ± 0.02 1.4 ± 0.07 22.5 ± 1.1 dispersion did not translate into better CO2 conversion.
The results observed in figure 3 can be better understood
and explained by the modification of the dielectric prop-
in coherence with the observed CO2 conversion. Nevertheless, erties of the medium between the two electrodes after the
after ZnO addition, both surface areas decreased, while CO2 introduction of a solid (either support or catalyst). After the
conversion decreased only for Cu/A. introduction of a solid into the DBD reactor, the breakdown
Along with the surface area, a question arises concerning voltage usually decreases as the discharge becomes easier to
the generation (and potential influence) of plasma in the cata- ignite [61]. Surface discharges around the solid pellets start to
lyst pores. It has been claimed that in order to generate plasma appear [62], increasing the plasma density and charge transfer
within the catalyst pores, the Debye length should be lower between electrodes along their gap.
than the pore size, as revealed by particle in cell + Monte The CO2 conversion was enhanced after the introduction of
Carlo collision model simulations [54]. The Debye length can the alumina support into the DBD reactor. This suggests that
be estimated through the following equation: the positive effect brought about by the increased dielectric
constant (because of the nonconductive character of the alu-
r
ε0 × kB × Te mina, table 6) is greater than the negative effect related to the
λD = , reduction of the residence time. After copper addition, CO2
n × e2
conversion decreased for the alumina support. This is related
where λD is the Debye length, ε0 is the permittivity of free to the two static dielectric constants (εS1 and εS2 ), which also
space (8.85 × 10−12 F m−1 ), kB is the Boltzmann con- decreased. Therefore, the addition of Cu decreased the energy
stant (1.38 × 10−23 J K−1 ), e is the charge of an electron storage capacity of the alumina leading to a decrease in the
(1.6 × 10−19 C), T e is the temperature of the electrons (K) plasma density and electron rates, which explains the lower
and n is the density of electrons (m−3 ). In our plasma DBD CO2 dissociation rates in the gas phase. Finally, the addition of
system, T e is around 3 eV (34 800 K) and the density of elec- ZnO slightly increased the static dielectric constants of Cu/A.
trons is approximately equal to 1018 m−3 [55, 56]. In this case, Nevertheless, this did not increase CO2 conversion as expec-
we would have a Debye length of 13 µm, which represents ted. The observed decrease in CO2 conversion might be related
three orders of magnitude larger than all catalysts’ pores in to a decrease in the catalytic contribution of CuZn/A towards
this study (table 3). From this point of view, plasma cannot be CO2 conversion due to its lowered surface area and increased
created inside the catalyst pores. However, we could consider copper reducibility (tables 3 and 4, respectively).
the diffusion of relatively long lifetime species inside them. In The modification of the dielectric properties also explains
the latter case, the internal surface area could be partially used the trends of the Z23 catalysts. The Z23 support presented
in plasma catalysis. a similar CO2 conversion to the empty reactor in figure 3.
For pores with smaller diameters than the Debye length, This implies that the increase in plasma density brought by
the plasma can only penetrate to some extent and at very the introduction of the Z23 was possibly countered by another
short times, i.e. at the beginning of a micro-discharge, before phenomenon. It was observed in section 3.2 that the Z23 sup-
the actual plasma streamer reaches the catalyst surface and port presented ionic conductivity at low frequency. A con-
a sheath is formed in front of the surface [54]. Zhang and ductive support would reduce the magnitude of the electric
Bogaerts studied the effect of surface charging on the forma- field (and ultimately the density of the plasma) as previ-
tion of plasmas inside catalyst pores for small and large pores. ously suggested by [29], thus explaining why CO2 conver-
They concluded that when the pore size is lower than 50 nm sion remained the same. The introduction of Cu into the
(which corresponds to this work), plasma cannot be formed Z23 support slightly increased both static dielectric constants,
inside the pores since plasma streamers at the dielectric surface but it also strongly enhanced its ionic conductivity (+39%).
will create sheath-rejecting plasma and only some short-lived This undesired increase in conductivity is responsible for the
reactive species can diffuse inside the pores. observed decrease in CO2 conversion. Finally, CO2 conver-
On the other hand, the TPR tests present an interesting trend sion was significantly increased for the Cu/Z23 after ZnO
between CO2 conversion and copper reducibility. Cu/Z23 addition. The addition of ZnO did not modify the two static
presented a higher H2 uptake (and thus copper reducibility) dielectric constants, but decreased the ionic conductivity of
than Cu/A, yet a lower CO2 conversion. After ZnO addition, Cu/Z23 (table 6), improving CO2 conversion.
the H2 uptake of the Cu/A increased, while that of the Cu/Z23 In summary, it can be concluded that the presence of a
decreased. Their corresponding CO2 conversions decreased solid in a DBD reactor can increase the density of the plasma

12
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

and the electron rates, but only if the ionic conductivity of the highest CO2 conversion (34%) mainly through its dissociation
introduced dielectric material is null or low. A denser plasma into CO, justifying the highest carbon monoxide yield. On the
enhances CO2 dissociation into CO, mainly by CO2 electron other hand, the CuZn/Z23 shows the highest methanol yield
impact dissociation, but with a minor contribution from elec- (0.5%), which means that it is the most suitable for CO2 hydro-
tron impact vibrational excitation. genation into methanol. This is mainly attributed to its greater
energy storage capacity coupled with the catalytic effects of
the addition of copper and zinc oxide. Concerning the meth-
4.2. Product selectivity
ane formation, it was thought to occur predominantly on the
First, let us start by addressing the observed methane and surface of the high-voltage electrode and thus the addition
methanol selectivity within the empty reactor. This could be of any catalyst seems to decrease it slightly, probably due to
related to the inner electrode of the reactor, which is mainly the reduced residence time of the gas. A comparison between
made of iron. Iron is an active metal for CO2 methanation [2] the results of the alumina support and those of the alumina-
and, to a much lower extent, methanol formation [31]. The sur- supported catalysts shows that the methanol yield remained
face sites present over the inner electrode could then catalyze constant while CO2 conversion decreased. This implies that
their formation, explaining their presence. With respect to the some of the methane produced by both Cu/A and CuZn/A were
supports, methane and methanol can only be produced over catalyzed to a certain degree by the copper sites. This also jus-
the metallic surface of the inner electrode since neither the tifies the fact that the methane yield remained similar between
alumina nor the Z23 support contains any metallic particles. Z23 and Cu/Z23 while CO2 conversion decreased.
As a consequence, the production of both methane and meth- Although these methanol and methane yields may appear
anol could have only been decreased by two possible causes: low, it is necessary to recall the operating conditions under
either a diminution of the reactants’ residence time (provoked which they were obtained. Below 90 ◦ C and at atmospheric
by the introduction of the respective supports), a difference in pressure, neither of them was detected by conventional cata-
the distribution of the produced species (due to the modified lysis, which opens up the prospect for CO2 hydrogenation
dielectric properties) or both. using DBD plasma under these operating conditions.
The increase in the methanol selectivity after copper addi-
tion to both supports implies that methanol formation is pos-
itively affected by the presence of copper sites (Cu0 and Cu+ , 5. Conclusion
see section 3.2.3) of the catalyst [1, 9–12]. Nevertheless, we do
not have enough information to be able to discuss the corres- In this work, the hydrogenation of CO2 into methanol was
ponding reaction mechanism. In the case of methane selectiv- proved to occur under atmospheric pressure and temperature
ity, its increase could be related to the improved availability below 90 ◦ C in a catalytic DBD plasma reactor, while the reac-
of dissociated hydrogen atoms over the copper surface, thus tion was found not to take place under these same conditions
favouring hydrogenation reactions. via conventional heating. Four catalysts using two different
It has been reported in the literature that zinc oxide addi- supports (alumina and ZSM-5), copper as the active metal and
tion into copper catalysts increases their selectivity towards ZnO as the promoter were prepared, fully characterized and
methanol [30, 63]. Even though most of these catalytic tests tested on CO2 hydrogenation under DBD plasma.
were performed under conventional heating, the reported find- CO2 was mainly dissociated into CO in the gas phase by
ings are in accordance with the experimental trend observed in both vibrational excitation and electron impact. CO2 dissoci-
this work; an increase in methanol selectivity after ZnO addi- ation was found to be closely related to the conductive and
tion. However, additional studies are required to identify the dielectric properties of the medium. The introduction of a solid
reaction mechanism under DBD plasma. Michiels et al [64] (either support or catalyst) within the DBD reactor led to modi-
recently published a micro-kinetic modelling study of plasma- fication of the aforementioned properties. An increase in the
assisted methanol formation by CO2 hydrogenation on a Cu dielectric constant enhances the magnitude of the electric field,
catalyst surface, in which the mechanisms of plasma catalysis thus enhancing the plasma density and favouring CO2 dissoci-
were compared with those of thermal catalysis. Their model ation. This effect was observed while using the alumina sup-
revealed that the reverse water-gas shift reaction followed by port. However, when the introduced solid is conductive, the
CO hydrogenation, together with the so-called formate path, magnitude of the associated electric field is decreased and CO2
mainly contributed to methanol formation in thermal catalysis, conversion is not improved, as observed with the Z23 support.
and that adding plasma generated radicals and intermediates The introduction of copper increased both methanol and
resulted in a higher methanol turnover frequency. In addition, methane production by catalytic effect, but it also considerably
the conversion of plasma-generated CO to HCO∗ and sub- modified the dielectric behaviour of the catalyst and thus CO2
sequent HCOO∗ or H2 CO∗ formation seemed to contribute to conversion. The Cu/A presented a decrease in CO2 conversion
methanol formation. Hence, the model suggested the import- due to a decrease in its dielectric constant, while the Cu/Z23
ant role of CO, but also O, OH and H radicals, as they influence also presented a decrease in CO2 conversion but because of an
the reactions that consume CO2 and CO. increase in its ionic conductivity.
Two catalysts prove to be the most interesting when determ- The introduction of zinc oxide increased both methanol and
ining which one performs best in DBD plasma. On the one methane selectivity by improving the copper reducibility and
hand, the alumina support makes it possible to obtain the modifying the Cu2+ /Cu+ ratio of the catalysts. It also modified

13
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

the dielectric properties of the solids. While methanol (and to improvement in the gas phase methanol-to-dimethylether
a lesser extent methane) formation was found to mostly take reaction Catal. Today. 337 195–200
place on the catalytic surface, the increased product selectivity [9] Sheldon D 2017 Methanol production - a technical history
Johns. Matthey Technol. Rev. 61 172–82
was attributed to a higher hydrogen availability at the Cu–ZnO [10] Agrell J, Birgersson H, Boutonnet M, Melián-Cabrera I,
interface. The ionic conductivity ofCu/Z23 decreased signi- Navarro R M and Fierro J L G 2003 Production of hydrogen
ficantly after ZnO addition, thus improving CO2 conversion. from methanol over Cu/ZnO catalysts promoted by ZrO2
Overall, the alumina support showed the highest CO2 conver- and Al2O3 J. Catal. 219 389–403
sion (34%) into CO since it did not present any ionic conduct- [11] Ahouari H, Soualah A, Le Valant A, Pinard L and Pouilloux Y
2015 Hydrogenation of CO2 into hydrocarbons over
ivity. On the other hand, the CuZn/Z23 showed the highest bifunctional system Cu–ZnO/Al2 O3 + HZSM-5: effect of
methanol yield (0.5%) because of its greater energy storage proximity between the acidic and methanol synthesis sites
capacity coupled with the catalytic effects of the addition of C. R. Chim. 18 1264–9
copper and zinc oxide. [12] Saito M, Fujitani T, Takeuchi M and Watanabe T 1996
Development of copper/zinc oxide-based multicomponent
catalysts for methanol synthesis from carbon dioxide and
Data availability statement hydrogen Appl. Catal. Gen. 138 311–8
[13] Valença G, Vieira R and Braga N 2015 Water gas shift reaction
on copper catalysts supported on alumina and carbon
All data that support the findings of this study are included nanofibers Chem. Eng. Trans. 43 931–6
within the article (and any supplementary files). [14] Gao J et al 2012 A thermodynamic analysis of methanation
reactions of carbon oxides for the production of synthetic
natural gas RSC Adv. 2 2358–68
Acknowledgments [15] Snoeckx R and Bogaerts A 2017 Plasma technology—a novel
solution for CO2 conversion? Chem. Soc. Rev.
46 5805–63
The authors gratefully acknowledge the project GENCOMM- [16] Bill A, Eliasson B, Kogelschatz U and Zhou L-M 1998
NWE334 for financial support. The project GENCOMM Comparison of CO2 hydrogenation in a catalytic reactor and
has been funded by the European Union with the European in a dielectric-barrier discharge Studies in Surface Science
Regional Development Fund (ERDF) through the Interreg and Catalysis vol 114, ed T Inui, M Anpo, K Izui,
S Yanagida and T Yamaguchi (Amsterdam: Elsevier)
North-West Europe Program. The authors also thank the LCS pp 541–4
laboratory technicians Benjamin Foucault, Marie Desmurs- [17] Eliasson B, Kogelschatz U, Xue B and Zhou L-M 1998
Lozier and Valérie Ruaux for their invaluable assistance. Hydrogenation of carbon dioxide to methanol with a
discharge-activated catalyst Ind. Eng. Chem. Res.
37 3350–7
ORCID iDs [18] Wang L, Yi Y, Guo H and Tu X 2018 Atmospheric pressure
and room temperature synthesis of methanol through
Alain Ledoux  https://orcid.org/0000-0002-1117-1142 plasma-catalytic hydrogenation of CO2 ACS Catal.
8 Art. no. 1
Federico Azzolina-Jury  https://orcid.org/0000-0003- [19] Men Y-L et al 2019 Highly dispersed Pt-based catalysts for
3807-8586 selective CO2 hydrogenation to methanol at atmospheric
pressure Chem. Eng. Sci. 200 167–75
[20] Neyts E C and Bogaerts A 2014 Understanding plasma
References catalysis through modelling and simulation—a review
J. Phys. D: Appl. Phys. 47 224010
[1] Jadhav S G, Vaidya P D, Bhanage B M and Joshi J B 2014 [21] Bogaerts A et al 2020 The 2020 plasma catalysis roadmap
Catalytic carbon dioxide hydrogenation to methanol: a J. Phys. D: Appl. Phys. 53 443001
review of recent studies Chem. Eng. Res. Des. 92 Art. no. 11 [22] Butterworth T, Elder R and Allen R 2016 Effects of particle
[2] De˛bek R, Azzolina-Jury F, Travert A and Maugé F 2019 A size on CO2 reduction and discharge characteristics in a
review on plasma-catalytic methanation of carbon packed bed plasma reactor Chem. Eng. J. 293 55–67
dioxide—looking for an efficient catalyst Renew. Sustain. [23] Butterworth T and Allen R W K 2017 Plasma-catalyst
Energy Rev. 116 109427 interaction studied in a single pellet DBD reactor: dielectric
[3] Tans P P and Keeling R 2020 Trends in Atmospheric Carbon constant effect on plasma dynamics Plasma Sources Sci.
Dioxide, National Oceanic and Atmospheric Administration Technol. 26 065008
(available at: www.esrl.noaa.gov/gmd/ccgg/trends) [24] Laer K V and Bogaerts A 2017 How bead size and dielectric
[4] Ha-Duong M, Gaultier M and deGuillebon B 2011 Social constant affect the plasma behaviour in a packed bed
aspects of Total’s Lacq CO2 capture, transport and storage plasma reactor: a modelling study Plasma Sources Sci.
pilot project Energy Proc. 4 6263–72 Technol. 26 085007
[5] Sempuga B C and Yao Y 2017 CO2 hydrogenation from a [25] Song Y et al 2017 ZSM-5 extrudates modified with
process synthesis perspective: setting up process targets phosphorus as a super effective MTP catalyst: impact of the
J. CO2 Util. 20 34–42 acidity on binder Fuel Process. Technol. 168 105–15
[6] Centi G and Perathoner S 2009 Opportunities and prospects in [26] Mei D, Zhu X, He Y-L, Yan J D and Tu X 2014
the chemical recycling of carbon dioxide to fuels Catal. Plasma-assisted conversion of CO2 in a dielectric barrier
Today 148 191–205 discharge reactor: understanding the effect of packing
[7] Olah G A, Goeppert A and Prakash G K S 2009 Beyond Oil materials Plasma Sources Sci. Technol. 24 015011
and Gas: The Methanol Economy (New York: Wiley) [27] Wang W, Kim H-H, van Laer K and Bogaerts A 2018 Streamer
[8] Schnee J, Quezada M, Norosoa O and Azzolina-Jury F 2019 propagation in a packed bed plasma reactor for plasma
ZSM-5 surface modification by plasma for catalytic activity catalysis applications Chem. Eng. J. 334 2467–79

14
J. Phys. D: Appl. Phys. 54 (2021) 334003 M Q Feliz et al

[28] Michielsen I et al 2017 CO2 dissociation in a packed bed DBD [47] Padley M B, Rochester C H, Hutchings G J and King F 1994
reactor: first steps towards a better understanding of plasma FTIR spectroscopic study of thiophene, SO2 , and CO
catalysis Chem. Eng. J. 326 477–88 adsorption on Cu/Al2 O3 catalysts J. Catal.
[29] Jo S, Kim T, Lee D H, Kang W S and Song Y-H 2014 Effect of 148 438–52
the electric conductivity of a catalyst on methane activation [48] Dulaurent O, Courtois X, Perrichon V and Bianchi D 2000
in a dielectric barrier discharge reactor Plasma Chem. Heats of adsorption of CO on a Cu/Al2 O3 catalyst using
Plasma Process. 34 175–86 FTIR spectroscopy at high temperatures and under
[30] Tagawa T, Nomura N, Shimakage M and Goto S 1995 Effect adsorption equilibrium conditions J. Phys. Chem. B
of supports on copper catalysts for methanol synthesis from 104 6001–11
CO2 + H2 Res. Chem. Intermed. 21 193–202 [49] Zdravkova V, Drenchev N, Chakarova K, Mihaylov M and
[31] Sriakkarin C, Umchoo W, Donphai W, Poo-arporn Y and Hadjiivanoc K 2016 FTIR study of CO and NO
Chareonpanich M 2017 Sustainable production of methanol coadsorption on Cu-ZSM-5: does a Cu+(CO)(NO)
from CO2 over 10Cu-10Fe/ZSM-5 catalyst in a magnetic complex exist? Bentham Sci. Publ. 6 152–61
field-assisted packed bed reactor Catal. Today. 314 114–21 [50] Schumann J, Kröhnert J, Frei E, Schlögl R and Trunschke A
[32] Koh M K, Wong Y J, Chai S P and Mohamed A R 2018 2017 IR-spectroscopic study on the interface of Cu-based
Carbon dioxide hydrogenation to methanol over methanol synthesis catalysts: evidence for the formation of
multi-functional catalyst: effects of reactants adsorption and a ZnO overlayer Top. Catal. 60 1735–43
metal-oxide(s) interfacial area J. Ind. Eng. Chem. 62 156–65 [51] Kohler M A, Cant N W, Wainwright M S and Trimm D L 1989
[33] Bacariza M C, Graça I, Lopes J M and Henriques C 2019 Infrared spectroscopic studies of carbon monoxide
Tuning zeolite properties towards CO2 methanation: an adsorbed on a series of silica-supported copper catalysts in
overview ChemCatChem 11 2388–400 different oxidation states J. Catal. 117 188–201
[34] Marco D D et al 2016 Dielectric properties of pure alumina [52] Azzolina Jury F, Polaert I, Estel L and Pierella L B 2013
from 8GHz to 73GHz J. Eur. Ceram. Soc. 36 3355–61 Synthesis and characterization of MEL and FAU zeolites
[35] Mir M A et al 2017 Utilization of zeolite/polymer composites doped with transition metals for their application to the fine
for gas sensing: a review Sens. Actuators B Chem. chemistry under microwave irradiation Appl. Catal. Gen.
242 1007–20 453 92–101
[36] Azzolina-Jury F 2019 Novel boehmite transformation into [53] Lopes A C, Costa C M, I Serra R S, Neves I C, Ribelles J L G
γ-alumina and preparation of efficient nickel base alumina and Lanceros-Méndez S 2013 Dielectric relaxation, ac
porous extrudates for plasma-assisted CO2 methanation conductivity and electric modulus in poly(vinylidene
J. Ind. Eng. Chem. 71 410–24 fluoride)/NaY zeolite composites Solid State Ion
[37] Quezada M, Azzolina Jury F, Polaert I, Ledoux A and 235 42–50
Fernandez C 2019 CO2 hydrogenation to methanol using [54] Zhang Q-Z and Bogaerts A 2018 Propagation of a plasma
Cu/ZSM-5 extrudates in a cold DBD plasma Presented at streamer in catalyst pores Plasma Sources Sci. Technol.
the 12th European Congress on Chemical Engineering 27 035009
(Florence, Italy) [55] Ronda-Lloret M, Wang Y, Oulego P, Rothenberg G, Tu X and
[38] Quezada M 2020 Hydrogénation catalytique de CO2 en Shiju N R 2020 CO2 hydrogenation at atmospheric pressure
méthanol en lit fixe sous chauffage conventionnel et sous and low temperature using plasma-enhanced catalysis over
plasma à DBD, INSA Rouen Normandie supported cobalt oxide catalysts ACS Sustain. Chem. Eng.
[39] Polaert I, Bastien S, Legras B, Estel L and Braidy N 2015 8 17397–407
Dielectric and magnetic properties of NiFe2 O4 at 2.45GHz [56] De Bie C, van Dijk J and Bogaerts A 2016 CO2 hydrogenation
and heating capacity for potential uses under microwaves in a dielectric barrier discharge plasma revealed J. Phys.
J. Magn. Magn. Mater. 374 731–9 Chem. C 120 25210–24
[40] Legras B, Polaert I, Estel L and Thomas M 2011 Mechanisms [57] Arena F, Italiano G, Barbera K, Bonura G, Spadaro L and
responsible for dielectric properties of various faujasites Frusteri F 2009 Basic evidences for methanol-synthesis
and linde type A zeolites in the microwave frequency range catalyst design Catal. Today 143 Art. no. 1
J. Phys. Chem. C 115 3090–8 [58] Dong X, Li F, Zhao N, Xiao F, Wang J and Tan Y 2016 CO2
[41] Polaert I, Benamara N, Tao J, Vuong T-H, Ferrato M and hydrogenation to methanol over Cu/ZnO/ZrO2 catalysts
Estel L 2017 Dielectric properties measurement methods prepared by precipitation-reduction method Appl. Catal. B
for solids of high permittivities under microwave 191 8–17
frequencies and between 20 and 250 ◦ C Chem. Eng. [59] Lei H, Hou Z and Xie J 2016 Hydrogenation of CO2 to
Process. 122 339–45 CH3 OH over CuO/ZnO/Al2 O3 catalysts prepared via a
[42] Boumaza S et al 2010 Water gas shift reaction over the solvent-free routine Fuel 164 191–8
CuB2 O4 spinel catalysts React. Kinet. Mech. Catal. [60] Saito M et al 1995 Development of Cu/ZnO-based high
100 145–51 performance catalysts for methanol synthesis by CO2
[43] Natesakhawat S et al 2012 Active sites and structure–activity hydrogenation Energy Convers. Manage.
relationships of copper-based catalysts for carbon dioxide 36 577–80
hydrogenation to methanol ACS Catal. 2 Art. no. 8 [61] Tu X, Gallon H J, Twigg M V, Gorry P A and Whitehead J C
[44] Gesmanee S and Koo-amornpattana W 2017 Catalytic 2011 Dry reforming of methane over a Ni/Al2 O3 catalyst in
hydrogenation of CO2 for methanol production in fixed-bed a coaxial dielectric barrier discharge reactor J. Phys. D:
reactor using Cu-Zn supported on gamma-Al2 O3 Energy Appl. Phys. 44 274007
Proc. 138 739–44 [62] Tu X, Gallon H J and Whitehead J C 2011 Transition behavior
[45] Meliancabrera I, Granados M L and Fierro J L G 2002 of packed-bed dielectric barrier discharge in argon IEEE
Pd-modified Cu–Zn catalysts for methanol synthesis from Trans. Plasma Sci. 39 2172–3
CO2 /H2 mixtures: catalytic structures and performance [63] Nakamura J, Choi Y and Fujitani T 2003 On the issue of the
J. Catal. 210 285–94 active site and the role of ZnO in Cu/ZnO methanol
[46] Pestryakov A N, Lunin V V, Kochubey D I, Yu. Stakheev A synthesis catalysts Top. Catal. 22 277–85
and Smolentseva E 2005 Influence of modifying additives [64] Michiels R, Engelmann Y and Bogaerts A 2020 Plasma
on formation of supported copper nanoparticles Eur. Phys. catalysis for CO2 hydrogenation: unlocking new pathways
J. D 34 55–8 toward CH3 OH J. Phys. Chem. C 124 25859–72

15

You might also like