Introduction

Global warming is a serious environmental issue that is exacerbated by the greenhouse effect. Mankind can work towards reducing this effect by decreasing the emission of anthropogenic greenhouse gases. Of these, CO2 from fuel combustion is by far the largest, accounting for 72% of the total emissions [1]. According to the analysis by the National Oceanic and Atmospheric Administration of the USA in 2016, the global annual temperature has increased at an average rate of 0.07 °C per decade since 1880 and at an average rate of 0.17 °C per decade since 1970 [2]. The atmospheric concentration of CO2 keeps an upward tendency from 370 ppm in 2000 to 405 ppm in 2018 [3]. To control the concentration of CO2 in the air, CO2 capture and utilization could be one of the attractive solutions for the mitigation of CO2 emissions.

Normally, hydrogen is used as co-reactant in the conventional CO2 to fuel synthesis. H2O is not only a cheaper hydrogen source compared with H2 and CH4, but also the common waste emitted with CO2 in industrial processes such as ammonia production. The direct conversion of CO2 and H2O is a promising approach based on the use of cheap, abundantly available raw materials.

$${\text{CO}}_{{2\left( {\text{g}} \right)}} + {\text{H}}_{2} {\text{O}}_{{\left( {\text{g}} \right)}} = {\text{CO}}_{{\left( {\text{g}} \right)}} + {\text{O}}_{{2\left( {\text{g}} \right)}} + {\text{H}}_{{2\left( {\text{g}} \right)}} \quad \Delta {\text{H}}_{\text{r}}^{\text{o}} = 525\,{\text{kJ/mol}}$$
(1)
$${\text{CO}} + {\text{H}}_{2} {\text{O }} \rightleftharpoons {\text{CO}}_{2} + {\text{H}}_{2} \quad \Delta {\text{H}}_{\text{r}}^{\text{o}} = - 41\,{\text{kJ/mol}}$$
(2)

However, converting CO2 and H2O incurs a large thermodynamic penalty, requiring high temperature to break chemical bonds (1) due to the chemical stability of both components. This may be followed by a Water Gas Shift Reaction (WGSR) as shown in (2). Even at 2000 K, CO2 conversion is only 1.5% [4]. Existing approaches for CO2 and H2O conversion include electrolysis and photocatalysis, but they have various shortcomings such as high operation temperature in electrolysis [5] and low efficiency of solar energy utilization in photocatalysis [6]. Here, we turn to non-thermal plasma, which can be generated by electrical discharges at atmospheric pressure. It provides a way to convert CO2 and H2O to syngas or hydrocarbons at low temperature and atmospheric pressure. Moreover, non-thermal plasma can be powered by electricity generated from renewable sources such as solar and wind, incorporating the CO2 molecules into a renewable carbon cycle that can reduce the dependence on fossil fuels. In addition, this method is suitable decentralized and small-scale CO2 conversion, which can be coupled to various CO2 sources at suitable locations.

The conversion of CO2 and H2O by non-thermal plasma was studied in DBD reactors [7, 8], microwave discharge reactors [9, 10], gliding arc discharge reactors [11], surface discharge reactors [12] and ferroelectric pellets packed-bed reactors [13]. However, the splitting of CO2 and H2O molecules remains a challenge. The main product reported in those researches is syngas with different H2: CO ratios. Methane [8] and dimethyl ether [12] have also been observed, but not quantified. Guo et al. [9] studied the simultaneous dissociation of CO2 and H2O in a surface wave sustained discharge operating at 915 MHz in a pulse regime, the highest yields of CO and H2 are lower than 8% and 4%, respectively, at a flow rate of 6 slm. They added NiO/TiO2 treated with Ar plasma in the CO2 and H2O reaction to explore the synergy effect between plasma and catalysts. The conversion of CO2 was improved from 23% (with plasma only) to 43%, with syngas as the sole product.

DBD is the discharge generated between two electrodes separated with an insulating barrier, which is widely applied in CO2 conversion by non-thermal plasma [14]. Catalysts can be packed into the discharge zone of the DBD reactor and work with plasma synergistically. Ramses et al. [7] studied the conversion of CO2 and H2O with H2O concentration in the feed gas from 0 to 8% in a DBD reactor. The conversions of CO2 and H2O were < 5% and syngas was produced with various H2/CO ratios up to 0.18. The ratio increased linearly with the increase of water content. Shaik et al. [8] reported that the CO2 conversion ascends from 24% with plasma to 36% when combining the plasma with Ni/Al2O3 catalysts. Carbon nanofibers were obtained in the packing region of the DBD reactor and CH4 was detected besides syngas in the product stream. However, the water conversion and the amount of methane produced were not mentioned in detail.

Our main aim here is to evaluate whether the addition of Ar and catalysts can increase the syngas ratios (H2:CO) and the conversions of CO2 and H2O. First, CO2 and H2O with various feed gas compositions were converted to syngas directly in the DBD reactor. Then, to improve the syngas ratio and H2O conversion, different contents of Ar were added in the feed gases for the purpose of avoiding the recombination of H2 and O2. Finally, Ni/γ–Al2O3 catalysts were combined with plasma to improve the syngas ratios of the product stream and obtaining CH4.

Experimental

Experimental Setup

The experiments are carried out in a cylindrical DBD reactor as shown in Fig. 1. A stainless-steel mesh (ground electrode) is wrapped over the outside of a quartz tube with outer and inner diameters of 13 mm and 10 mm, respectively (dielectric thickness = 1.5 mm). A stainless-steel rod with an outer diameter of 8 mm is placed in the center of the quartz tube and it acts as an inner electrode. The length of the discharge region is 100 mm with a discharge gap of 1 mm, which resulted in a discharge volume of 2.83 cm3. The inner electrode is connected to a high voltage output and the outer electrode is grounded via an external capacitor (100 nF). The DBD reactor is supplied by an AC high voltage–power supply (AFS generator G15S-150 K) for a maximum power of 1500 W, with a maximum peak-to-peak voltage of 60 kV and a frequency range of 10–150 kHz. The applied voltage (V1) is measured by a high voltage probe, while the voltage (V2) on the external capacitor is measured by another voltage probe. All of the electrical signals are measured by a four-channel digital oscilloscope. The Q–U Lissajous method is used to calculate the discharge power. The specific energy input (SEI) is equal to the ratio of the calculated discharge power to the gas flow, as in (3). The output products are detected by Gas Chromatography (Thermal Scientific Trace 1300) directly with Flame Ionization Detector (FID) for CH4 and Thermal Conductivity Detector (TCD) for other gases.

Fig. 1
figure 1

Schematic diagram of the experimental setup

$${\text{SEI }}\left[ {{\text{J cm}}^{ - 3} } \right] = \frac{{{\text{discharge power }}\left[ {\text{W}} \right]}}{{{\text{flow rate }}\left[ {{\text{ml min}}^{ - 1} } \right]}} \cdot \frac{{60\left[ {{\text{s min}}^{ - 1} } \right]}}{{1\left[ {{\text{cm}}^{3} {\text{ml}}^{ - 1} } \right]}}$$
(3)

In the plasma catalytic section, the catalysts were fully packed into the discharge volume within the 1 mm gap. The reactions with CO2 and H2O were carried out using plasma alone, plasma with support (Al2O3, TiO2) and plasma with catalysts (Ni/Al2O3) in the DBD reactor. The feeding gases, CO2 and Ar were provided from gas cylinders and controlled by a set of mass flow controllers. H2O was evaporated and mixed with CO2 in a controlled evaporator mixer. The gas line and the reactor were heated to 105 °C in order to avoid water condensation during reaction. The gas flow rates of CO2, H2O and Ar can be adjusted at ranges of 0–200 mL/min, 0–20 ml/min and 0–200 ml/min, respectively. The conversion, yield of CO2 and H2O were calculated by comparing the CO2 and H2O peak areas of the GC before and after plasma, processing based on following equations,

$${\text{CO}}_{2} {\text{conversion }}\left[ \% \right] = \frac{{{\text{moles of CO}}_{2} {\text{converted}}}}{{{\text{moles of CO}}_{2} {\text{in feed}}}} \cdot 100{\text{\% }}$$
(4)
$${\text{Yield of CO}}\left( {{\text{H}}_{2} } \right)\left[ \% \right] = \frac{{{\text{moles of CO}}\left( {{\text{H}}_{2} } \right) {\text{produced}}}}{{{\text{moles of CO}}_{2} \left( {{\text{H}}_{2} {\text{O}}} \right) {\text{in feed}}}} \cdot 100{\text{\% }}$$
(5)

To analyze the products, two different selectivities are defined, C-based selectivity for the C-containing species (e.g. CO, CH4) and the H-based selectivity for the H-containing species (e.g. CH4, H2).

H-based selectivity as Eqs. (6, 7):

$${\text{S}}_{{{\text{H}},{\text{H}}_{2} }} = \frac{{{\text{moles of H}}_{2} {\text{produced}}}}{{{\text{moles of H}}_{2} {\text{O consumed}}}}$$
(6)
$${\text{S}}_{{{\text{H}},{\text{CH}}_{4} }} = \frac{{{\text{moles of CH}}_{4} {\text{produced }}}}{{{\text{moles of H}}_{2} {\text{O consumed}}}} \cdot 2$$
(7)

C-based selectivity as Eqs. (8, 9):

$${\text{S}}_{{{\text{C}},{\text{CO}}}} = \frac{\text{moles of CO produced}}{{{\text{moles of CO}}_{2} {\text{consumed}}}}$$
(8)
$${\text{S}}_{{{\text{C}},{\text{CH}}_{4} }} = \frac{{{\text{moles of CH}}_{4} {\text{produced }}}}{{{\text{moles of CO}}_{2} {\text{consumed}}}}$$
(9)

Catalyst Preparation

Ni/γ–Al2O3 catalysts were prepared by wet impregnation, using nickel nitrate hexahydrate Ni(NO3)2·6H2O, 98.5%, Sigma-Aldrich) as a Ni precursor and commercial Al2O3 (Sigma-Aldrich) as support. The right amount of precursor was used to obtain 10, 20 and 30 wt% of metal loading. The formed slurry was stirred for 24 h at 60 °C, followed by 2-h drying at 120 °C and calcination in an open-air furnace at 450 °C for 4 h to obtain NiO/γ–Al2O3. Before the experiments, the catalysts were reduced in 20% H2/Ar flow with a flow rate of 30 ml/min at 700 °C for 1 h and are denoted as Ni/γ–Al2O3.

Characterization of the Catalysts

The following techniques were used to characterize the catalysts before and after the CO2 and H2O reactions in the plasma-catalytic reactor. The surface area of the catalysts was determined by N2 adsorption at a − 196 °C, using BET analysis. Temperature programmed reduction (TPR) was applied to check the reducibility of the catalysts. Before TPR measurements, the catalyst samples were heated in He at 300 °C for 1 h and flushed with Ar for 1 h at 50 °C. Afterwards, the sample was cooled down in flowing Ar to room temperature. The Ar flow was changed to 10% H2/Ar and the reactor was heated linearly at a rate of 10 °C/min up to 900 °C. The H2 consumption was detected by a thermal conductivity detector (TCD). Analysis of the crystalline structure of the catalyst was conducted by X-ray Diffraction (XRD) at a scanning rate of 0.02°/20 s from 2θ 10°–90°. The mass loss of the catalysts was measured by Thermogravimetric Analysis (TGA) in air by heating up to 1000 °C at a rate of 10 °C/min. Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM) were used to observe the particle size of the catalysts and the structure of the samples.

Results and Discussion

Effect of H2O Content on the Conversion of CO2 and H2O

The impact of the H2O content from 0 to 10% on the conversion of CO2 and H2O was examined at a total rate of 100 ml/min in a blank DBD reactor as shown in Fig. 2. The operated voltage and frequency were 14 kV (from peak to peak) and 30 kHz respectively. The value of SEI was 55.4 J/cm3 for all the reactions, which was calculated by integrating the area of Q–V Lissajous figure. As the concentration of H2O increased, the conversion of CO2 decreased from 8.7 to 2.9% and the conversion of H2O decreased slightly from 1.7 to 1.0%, as shown in Fig. 2.

Fig. 2
figure 2

CO2 and H2O conversion as a function of the water vapor content

The product compositions and syngas ratios are shown in Table 1. The main product is syngas with different H2/CO ratios. The addition of H2O led to a decline of the CO2 conversion and an increase of H2 production, resulting in a rise of syngas ratio in product, this is in line with reported literature [7, 11, 13]. In this way, the syngas ratio can be adjusted by changing the H2O/CO2 ratio in the feed gases to fit the processing needs of gas fermentation. However, the current H2O conversion is less than 2% and the syngas ratio is too low to be applied in the industrial production.

Table 1 Product composition after plasma treatment

There are two possible reasons for the low H2O conversion are as follows: (1) The addition of H2O vapor in the feed gas leads to a reduction in the number of micro-discharges [15] and electron density [7], resulting in low conversions of CO2 and H2O [11]. (2) Rapid recombination reactions in plasma limit both of CO2 and H2O feed gases conversions and syngas ratios in the product.

Rate coefficient k (cm3 molecule−1 s−1) [7]

$${\text{CO}}_{2} + {\text{e}}^{ - } \to {\text{CO}}_{{}} + {\text{O}}_{{}} + {\text{e}}^{ - } \quad { - }$$
(10)
$${\text{H}}_{2} {\text{O}} + {\text{e}}^{ - } \to {\text{OH}} + {\text{H}} + {\text{e}}^{ - } \quad { - }$$
(11)
$${\text{CO}} + {\text{OH}} \to {\text{CO}}_{2} + {\text{H}}\quad 2.8 \times 10^{ - 14}$$
(12)
$${\text{H}} + {\text{OH}} \to {\text{O}} + {\text{H}}_{2} \quad 2.1 \times 10^{ - 16}$$
(13)
$${\text{H}}_{2} + {\text{OH}} \to {\text{H}}_{2} {\text{O}} + {\text{H }}\quad 1.1 \times 10^{ - 14}$$
(14)
$${\text{O}} + {\text{O}} + {\text{M}} \to {\text{O}}_{2} + {\text{M}}\quad 7.0 \times 10^{ - 33}$$
(15)
$${\text{H}} + {\text{O}}_{2} + {\text{M}} \to {\text{HO}}_{2} + {\text{M }}\quad 1.4 \times 10^{ - 32} - 2.9 \times 10^{ - 31}$$
(16)
$${\text{HO}}_{2} + {\text{O}} \to {\text{O}}_{2} + {\text{OH}}\quad 5.4 \times 10^{ - 11}$$
(17)
$${\text{H}} + {\text{OH}} + {\text{M}} \to {\text{H}}_{2} {\text{O}} + {\text{M}}\quad 1.5 \times 10^{ - 30} - 2.2 \times 10^{ - 29}$$
(18)

The possible reaction path was listed as (1018), the rate coefficient of related reactions was reported in a study of zero-dimension kinetic model by Ramses [7]. The formation of CO and H2 are mainly through electron impact induced reaction (10) and (11). CO obtained from CO2 dissociation can recombine with OH to form CO2 rapidly and gives rise to the reduction of CO2 conversion (12). This equation shows a higher reaction rate than the following reactions like H2 production (13) and three body recombination of O2/H2O (15, 18), which was considered as the main reason leading to the low conversion of CO2 [7]. The H atoms produced in Eqs. (11, 12) react further to generate H2, OH, H2O as Eqs. (13, 17) and react with OH back to H2O through three body recombination process (18), resulting in the low H2O conversion. Other species such as O3 and H2O2 can also form in theory, but analysing these is out of the scope of this paper.

Effect of Ar on the Conversion of CO2 and H2O

The conversion of CO2 in DBD reactor rises dramatically with the addition of Ar [16]. The breakdown voltage is lower in the CO2/Ar mixtures owing to the higher Townsend ionization coefficient of Ar, this indicates that a larger fraction of the power can be utilized effectively for the CO2 dissociation, as less power will be dissipated for the gas breakdown [16]. The charge could be transferred from the Ar+ ions to CO2, which promotes the CO2 splitting as shown in (19, 20). However, the ionisation of Ar requires a much higher electron energy (15.76 eV) than that for the excitation of Ar such as 11.55 eV for Ar (4s3P2), and 11.72 eV for Ar (4 s3P0) [17]. Thus, Ar is more likely excited to its metastable state rather than being ionized as given in the (21). The presence of the metastable Ar species (Ar*) could create new reaction pathways for the dissociation of CO2 as in (22) [18]. As H2O conversion was lower than 2% in the tests (Sect. 3.1), the effect of Ar was explored in the reaction to dilute the species generated in the reaction and avoid the recombination reactions.

$${\text{Ar}}^{ + } + {\text{CO}}_{2} \to {\text{Ar}} + {\text{CO}}_{2}^{ + }$$
(19)
$${\text{CO}}_{2}^{ + } + {\text{e}}^{ - } \to {\text{CO}} + {\text{O}}$$
(20)
$${\text{e}}^{ - } + {\text{Ar}} \to {\text{e}}^{ - } + {\text{Ar}}^{*}$$
(21)
$${\text{Ar}}^{*} + {\text{CO}}_{2} \to {\text{Ar}} + {\text{CO}} + {\text{O}}$$
(22)

The impact of the addition of 10%, 50% and 90% of Ar in the feed gases on the conversion of CO2 and H2O was studied at a total rate of 100 ml/min in the DBD reactor as shown in Fig. 3. We could only test the H2O content up to 5% in 90% Ar, due to the limitation of the setup. The operated voltage and frequency were 14 kV (peak to peak) and 30 kHz, respectively. Generally, as the concentration of Ar increases, the conversion of H2O has an increasing tendency from 1 to 4.1%. However, the conversion of CO2 decreased from 8.7 to 3.1%. As for the tendency described in Sect. 3.1, the conversions of both CO2 and H2O showed a decreasing trend with the rise of H2O content. The products are syngas with various ratios. The values of SEI and syngas ratio are shown in Table 2.

Fig. 3
figure 3

The effect of Ar on the conversion of CO2 (a) and H2O (b) at different H2O contents

Table 2 Specific energy input (SEI) values and syngas ratios in the product stream

Adding Ar promoted the conversion of H2O, while the conversion of CO2 decreased simultaneously. The tendency is different from the effect of Ar on the conversion of CO2 alone [16]. The syngas ratios follow an upward tendency up to 0.86 with the increase of the H2O content. The dissociation energy of H2O (2.6 eV/mol) is slightly lower than that for the CO2 dissociation (\(2.9{\text{eV}}/{\text{mol}})\). Charge transfer between the reactive Ar species (e.g. Ar*, Ar+ ions) and H2O might contribute to the H2O dissociation. As a result, it promotes reverse reaction of CO2 leading to the decrease of the CO2 conversion. When the H2O content is increased, the conversions of both CO2 and H2O are decreased in all cases. The reason could be that the addition of H2O leads to a reduction in electron density as discussed in Sect. 3.1.

Catalytic Effect on the Conversion of CO2 and H2O in the DBD Reactor

As summarized in the above discussion, both H2O conversion and syngas ratio were enhanced with the addition of Ar. However, the conversion of CO2 and H2O are still lower than 10%; besides, syngas ratio is slightly lower than the one for acetic acid production by Fischer–Tropsch synthesis. To improve the conversion and obtain hydrocarbons directly from the reaction, catalysts were packed into DBD reactor. Supported Ni catalysts were proved to enhance the conversions in CO2–CH4 dry reforming reaction by non-thermal plasma [19,20,21,22]. It also was used to improve the selectivity of CH4 for photocatalytic CO2 reduction by H2O [23]. Thus, supported Ni catalysts were chosen as the catalysts in our experiments. The reaction with 90% Ar showed highest conversion of H2O. Lower H2O content supports the CO2 conversion, thus, the following experiments were performed with 90% Ar, 2% H2O and 8% CO2.

Flow Rate Tests for Reaction

To study the effect of the flow rate on the reaction, 70 ml/min (lowest flow rate of the setup), 100 ml/min, 150 ml/min and 200 ml/min were applied in the plasma reaction. The operated voltage and frequency were 14 kV (from peak to peak) and 30 kHz, respectively. The conversions of CO2 and H2O are shown in Fig. 4.

Fig. 4
figure 4

Conversion of CO2 and H2O with various flow rates

The conversions of feed gases showed a downward trend with a rise of total flow rate. The highest conversion of CO2 and H2O was obtained at the lowest flow rate of 70 ml/min of feed gas. It coincides with the results in the CO2 dissociation [24] and CO2 hydrogenation in DBD reactor [25]. Low residence time leads to the low conversion of feed gases. Therefore, the optimal/lowest flow rate of 70 ml/min was selected for plasma catalytic experiments.

Plasma Catalytic Reactions: Effect of γ-Al2O3 and TiO2

γ-Al2O3 and TiO2 are the most common catalyst supports in plasma-catalytic reactions, which are chosen to compare the effects on the reaction resulting from various dielectric constants and specific surface areas. The first plasma-catalytic experiments were performed with these two supports. Experiment with blank reactor was also performed for comparison. The conversion of CO2 and H2O as a function of time is presented in Fig. 5, and the Q–V Lissajous figures of these three cases are presented in Fig. 6. The operated voltage and frequency were 12 kV from peak to peak (the highest stable voltage for a long-term run) and 30 kHz in the reaction, respectively. Because Al2O3 has a strong water absorption ability, product measurements were taken after 15 min of plasma treatment. In this way, the adsorbed water is released before the measurement so that the measurement results will not be affected. The detected products were CO and H2. Figure 5 shows that TiO2 was effectively enhanced CO2 conversion, around 20–24% CO2 conversion was achieved, which is 15% higher than the case when no catalyst is used. However, TiO2 had no obvious effect on H2O conversion. The reaction with γ-Al2O3 promotes CO2 conversion from 5% (only plasma) to 10% and H2O conversion from 1.6 to 2.5%.

Fig. 5
figure 5

Conversion of CO2 (a) and H2O (b) using different supports

Fig. 6
figure 6

Lissajous figures of reaction in DBD reactor without and with different packing material

The shape of the Lissajous figure changes from parallelogram to oval shape when either TiO2 or Al2O3 are fully packed in the DBD reactor, resulting in the change of discharge mode from typical filamentary discharge to a combination of weak filamentary and predominantly surface discharge. The pellets fully packed into the discharge gap result in significant reduction in the discharge volume, limiting the distance where filamentary microdischarges can form [26]. As a result, less intense filament current peaks can be generated in the void between pellet–pellet and pellet–quartz wall as shown in Fig. S1. Meanwhile, a surface discharge can be formed on the surface of pellets near the contact points between the pellets, which is introduced as a packed-bed discharge effect [26, 27]. The breakdown voltage of the discharge does not change significantly with/without the packing materials from Fig. 6. The different discharge behaviors with packing pellets are caused by the enhanced local electric field strength near the contact points between the pellets and the pellet–dielectric wall [28]. From the measured V–I characteristics shown in Figure S1, differences in the number of microdischarges are observed with an order as: Blank > Al2O3 > TiO2. In addition, the SEI in the case of TiO2 is slightly higher than Al2O3 (29.3 J/cm3 and 25.8 J/cm3 respectively), which could also contribute to the difference in conversion of CO2. The difference in discharge characteristics between Al2O3 and TiO2 is not significant. We believe that the main reason for the difference in conversion is the physical properties of the packing material. The surface area of TiO2 is much smaller than Al2O3 (Table S3). Indeed the plasma generation in micropores is less likely to occur, however, Al2O3 with large surface area and strong H2O absorption capacity can absorb more H2O on the surface compared with TiO2 at the same flow rate, leading to higher residence time of H2O on the surface of pellets which promotes the H2 production. At the same time more OH is produced, which will react with CO, resulting in a lower conversion of CO2. On the other hand, the rate of CO reverse reactions is mainly determined by the surface rather the gas phase. Therefore, TiO2 with lower surface area gives a lower rate for CO to react with oxygen species, hence higher CO2 conversion is observed [29].

Since γ-Al2O3 showed the best H2O conversion results, it was used as the support for the following experiments. The introduction of Ni/γ–Al2O3 catalysts in the plasma discharge was reported to improve the performance of CO2 activation reactions with hydrocarbons or with H2O, in comparison with the plasma-only reaction [8, 30, 31]. Thus, Ni/γ–Al2O3 was used to explore the catalytic effect on this reaction further.

Plasma Catalytic Reactions: Effect of Ni/γ–Al2O3 on the Conversion of CO2 and H2O

To study the effect of different metal loadings on the catalytic activity, 10% Ni/Al2O3, 20% Ni/Al2O3 and 30% Ni/Al2O3 catalysts were studied and compared with blank experiments. The results are presented in Fig. 7.

Fig. 7
figure 7

Conversion of CO2 (a) and H2O (b) with no catalyst, Al2O3, 10% Ni/Al2O3, 20% Ni/Al2O3, 30% Ni/Al2O3 and 20% NiO/Al2O3

From Fig. 7, 10% Ni/Al2O3 significantly showed a positive effect on the conversion of CO2, converting between 10 and 14% of CO2 and showing the best CO2 conversion results compared with the others in 2-h reaction. It also has slightly higher conversion of H2O compared with the blank and Al2O3. CO2 conversion with 30% Ni/Al2O3 catalyst was similar to Al2O3, while H2O conversion was slightly higher. It is evident that 20% Ni/Al2O3 shows the lowest CO2 conversion values compared with other conditions. However, it is the most active for H2O conversion. Although the conversion declines from 14.7 to 4% after 120 min, these values are always higher than the H2O conversions in the other cases.

Furthermore, the products in all cases were syngas with different ratios, apart from 292 ppm CH4 obtained with 20% Ni/Al2O3. H2 and CO yields were up to 13.8% and 5.6%, respectively (Fig. 8). CH4 achieved the highest yield 0.32% at 15 min of reaction (first measurement) and it was not produced anymore after 75 min of reaction. The highest syngas ratio was 0.6 after 15 min due to the high yield of H2, and it decreased to 0.2 after 2 h of reaction. Thus, metallic Ni is more active than NiO for H2O conversion, and it can produce CH4 from Figs. 8 and 9.

Fig. 8
figure 8

Yield of valuable products in the reaction using 20% Ni/Al2O3 catalyst

Fig. 9
figure 9

Selectivity of valuable products in the reaction with 20% Ni/Al2O3

The C-based selectivity and H-based selectivity are depicted in Fig. 9. The C-based selectivity of CH4 decreased from 7 to 0% with time. Reversely, the selectivity of CO increased from 93 to 100% after 75 min of reaction. Besides, the maximum value of H-based CH4 selectivity was 11%, descending with reaction time.

The ability of Ni/Al2O3 to increase CO2 conversion was observed previously [8], which is in accordance with results of 10% Ni/Al2O3. However, Ni/Al2O3 catalysts are also active for the Water Gas Shift (WGS) reaction (2) [32], which could explain the catalytic behavior of 20% Ni/Al2O3 catalyst. The CO produced from CO2 dissociation could react with H2O, promoting the production of H2 and CO2 by the WGS reaction. However, the conversion decreased with time in 90 min. One possible explanation for this behavior is that CO dissociated to C (Boudouard reaction), reducing the conversion of H2O. This carbon can react with H2 further to produce CH4, as shown in Fig. 10 [18]. When the H2 concentration in the product mixture decreased, CH4 production decreased to zero in an hour. It was also reported that CO2 by in situ formation of hydrogen during water splitting with Ni catalysts may result in the production of CH4 and PCVD of CH4, leading to the formation of carbon fibers in a DBD reactor [8]. Because CH4 was detected during the 20% Ni/Al2O3 reaction, so the other possible reason is that the formation of carbon filaments from the PCVD of CH4 may reduce the catalyst activity and H2 production.

Fig. 10
figure 10

Reprinted with permission from [18], copyright 2017 Springer

Possible major reactions for the formation of CH4 on the Ni/Al2O3 catalyst.

To conclude, we have seen that inserting packing materials in the DBD plasma discharge can enhance the conversions compared with plasma alone. 10% Ni/Al2O3 shows the best CO2 conversion (14%) but less significant for H2O conversion. On the other hand, the reaction with 20% Ni/Al2O3 catalyst present much higher H2O conversion (14.7%) and lower CO2 conversion (5.9%) compared with 10% Ni/Al2O3, suggesting 20% Ni/Al2O3 is more effective for side reactions (WGS) and promotes H2 production. Further, the CO2 conversion of 20% Ni/Al2O3 reaction is lower than that in blank experiments, indicating that the WGS reaction occurs during 20% Ni/Al2O3 reactions and reduces CO production. Unlike 20% Ni/Al2O3, all the CO2 and H2O conversions in other cases were enhanced compared with blank experiments, suggesting that the dissociation reactions of CO2 and H2O to CO and H2 play a dominating role. Besides, it is also possible that Ni reacts with H2O to produce NiO and H2 (Ni + H2O = NiO + H2) during the reaction and increases the H2 production at the beginning of reaction with 20% Ni. The experiments with 30% Ni gave low conversions, suggesting the thermal catalysis and sintering may occur especially in the case of 30% Ni/Al2O3. Further characterization of the materials was performed to study the observed catalytic behavior.

Catalysts Characterization

X-ray Diffraction (XRD) Measurements

The XRD patterns for the 10% Ni/Al2O3, 20% Ni/Al2O3, 30% Ni/Al2O3 catalysts after reduction and after reaction are shown in Fig. S6. The characteristic peaks of NiO are not obvious in 10%/20% Ni and absent from the 30%Ni sample. Therefore, it cannot be confirmed that NiO formation is the dominating factor. The Ni peak of 30% Ni/Al2O3 after reaction is more intense than the peak before reaction unlike peaks of 10%/20% Ni/Al2O3. The average crystallite size of Ni in 30% Ni/Al2O3 catalyst was estimated by the Scherrer formula using the half width of the Ni (200) peak, which was 16.5 nm after reduction and 25.3 nm after reaction. This shows that sintering of Ni particles takes place during the reaction with 30% Ni/Al2O3.

Transmission Electron Microscopy (TEM) and Scanning Electron Microscopy (SEM) Analysis

TEM and SEM analysis of the catalysts after reduction (noted as before reaction) and after reaction were performed in order to study the formation of carbon deposits. Compared with the clean surface of reduced Ni/Al2O3 samples (before reaction) shown in Fig. S7. (a, c, e) from SEM images, filamentary deposits were clearly present on the surface of the catalysts after their use in the reaction for 2 h [Fig. S7. (b, d, f)]. This carbon can be identified as filamentary carbon [33, 34]. With TEM images, the amount of carbon filaments in 10% Ni/Al2O3 catalyst is small, making it difficult to identify. From Fig. S7 (a, b), the carbon filaments are not encapsulating the particles, therefore they do not influence the catalytic activity of this material. For 20% Ni/Al2O3, more carbon filaments are observed [Fig. S8 (c, d)]. They stay closer to the metal clusters, causing partial encapsulation. Fig. S8 (e, f) clearly shows that the carbon formed on the surface of 30% Ni/Al2O3 catalyst is encapsulating the metal particles. Curiously, those filaments form a V shape. Deactivation of the Ni catalysts by coking is a common problem [35, 36]. Coking can be produced from two different reactions: the Boudouard reaction (2CO → CO2 + C) and/or the hydrocarbon decomposition. It is possible that CH4 is produced, but it decomposes to C and H2.

In summary, the amount of metal loading strongly affects the properties of the final catalysts. 30% Ni/Al2O3 catalyst has bigger metal clusters with a weak interaction with the support (Fig. S5) and a significantly lower surface area than the other two materials (Table. S3), which may be the main reason for its poor catalytic activity. Besides, temperature gradients at the catalyst surface in DBD reactor may arise and influence the overall catalyst activity [37]. The heat capacity of metals Ni (Cp = 0.44 J g−1 K−1) is typically lower than the heat capacity of most commonly dielectric support materials Al2O3 (Cp = 0.72 J g−1 K−1) while the thermal conductivity of metals is much higher (κ = 90.9 W m−1 K−1 for Ni, κ = 20 W m−1 K−1 for Al2O3) [37]. Therefore, the temperature will be higher in the Ni surface in the plasma reaction leading to the sintering and deactivation of catalysts especially for 30% Ni/Al2O3. Moreover, it induces the formation of encapsulating carbon filaments that block the active sites. For 20% Ni/Al2O3 catalyst, it is the least active material for CO2 conversion, but the best catalyst for H2O conversion. This behavior may reflect the presence of side reactions such as WGS reaction, methane formation/PCVD and Ni oxidation as explained in Sect. 3.3.3. But Ni oxidation to NiO was not obvious from XRD, suggesting that reducing species such as hydrogen and CO produced in the plasma may react with NiO and reduce it. Among the studied materials, 10% metal loading appeared to be the best one for CO2 conversion, although it could not convert H2O. Characterization results (Fig. S5) showed that 10% Ni/Al2O3 catalyst contains highly dispersed NiO nanoparticles with a strong interaction with the support, giving a material with a high surface area. In this case the metal-support interface can adsorb more CO2, while the formation of carbon filaments is partially avoided.

Conclusions

The conversion of CO2 and H2O was carried out in a DBD plasma reactor at 105 °C. The effects of argon addition, packed support material and supported catalysts were studied. The addition of Ar produces reactive Ar species and reduces the recombination of H2 and O2, which promotes the H2O conversion. In this case, the syngas ratio can be adjusted to 0–0.86, which approaches the ratio for hydrocarbon synthesis via Fischer–Tropsch synthesis or gas fermentation. Further, the plasma catalytic CO2 and H2O reaction was explored with Ni/Al2O3 catalyst to improve the conversion of feed gases and to give access to hydrocarbons. Thus, the combination of plasma and a 10% Ni/Al2O3 catalyst facilitates the CO2 conversion. The 20% Ni/Al2O3 catalyst promotes the water–gas shift reaction (WGSR) and enhances the H2O conversion up to 14.7%. In addition, 292 ppm CH4 was obtained by the plasma reaction, which assumed to be formed from CO through dissociation and hydrogenation. The effect of the 30% Ni/Al2O3 catalyst is similar to Al2O3 as packing material which is accounted to sintering of the catalyst and carbon deposition.