Gliding Arc/Glow Discharge for CO₂ Conversion: The Role of Discharge Configuration and Gas Channel Thickness

Abstract

This work investigates CO₂ conversion using atmospheric pressure low-current gliding discharges (GD). The following three modifications are studied: classic GD; magnetically accelerated GD (MAGD); and magnetically retarded GD (MRGD). In the latter two, permanent magnets produce a magnetic field that either accelerates or retards the discharge downstream. The gas flow is confined between quartz plates and the electrodes, with varying channel thicknesses. The magnetic configurations improve the performance compared to the classic GD, with up to 30% higher energy efficiency and up to a 50% higher conversion rate. The highest conversion rate is 11–12% with 10% energy efficiency, while the highest efficiency is 40% with 5% conversion, achieved with MRGD and MAGD at channel thicknesses of 2 mm and 3 mm.

1. Introduction

The control and utilization of greenhouse gases is an integral part of the concept of systemic ecological thinking. As a result of industrial and other human activities, the most significant polluting gas is CO₂ [1]. One of the strategies to solve this problem is an activity called “Carbon Capture and Storage” (CCS), in which the polluting gas emitted from various sources is captured and isolated from the atmosphere. However, a more beneficial goal is to close the loop, recycle the CO₂, and reuse it as a raw material in industry. The technology developed is known in the literature as “Carbon Capture, Utilization and Storage” (CCUS) [2,3].

The first step before the production of value-added chemicals and fuels is to convert CO₂ into CO and O₂, which is very energetically expensive and requires the gas to be heated to at least a few thousand K (all CO₂ is converted at 5000 K). Various traditional techniques, such as pure thermal CO₂ splitting, conversion of CO₂ with co-reactants such as H₂ and CH₄, CO₂ + H₂O artificial photosynthesis and novel techniques such as photochemical, electrochemical, solar thermochemical, biochemical, and catalytic conversion, are being developed and researched [4]. In addition, plasma technology has been suggested and studied for CO₂ conversion [5,6,7]. A critical comparison of the different methods based on the main parameters—energy efficiency and conversion rate—shows that plasma technology is not the most optimal, but undeniably has four significant advantages. It is flexible in terms of energy consumption, plasma sources can be easily scaled, and they can be integrated in the industry at a reasonable cost [4,7]. Finally, they do not require any rare earth materials. These advantages of plasma technology are combined with a significant potential for increasing the energy efficiency and conversion rate of CO₂ into CO.

CO₂ is a thermodynamically stable triatomic molecule with the following four degrees of freedom for vibration: a symmetric stretch mode; two degenerate bending modes; and an asymmetric stretch mode [5]. Therefore, the stepwise dissociation via the vibrationally excited states of the molecules has been theoretically established as a reaction with considerable potential [8]. However, in arc plasmas, thermal dissociation with collisions between electrons and O atoms is usually the main mechanism for dissociation.

Three of the most widely studied discharges in this area are dielectric barrier discharges (DBDs), microwave (MW) discharges, and gliding arc discharges (GADs) [4]. The DBDs maintain a relatively high conversion (~30–40%), but with a very low energy efficiency of below 5%. MW discharges are known as devices with high energy efficiency and conversion rates at reduced pressure. For example, the measured values for a discharge operating at 915 MHz are 80% for energy efficiency and 35% for the conversion rate [5]. However, these results were achieved at low pressures, which requires an expensive vacuum system.

GADs, on the other hand, usually operate at atmospheric pressure, have a simple construction without a complicated and expensive vacuum system, and a comparatively high energy efficiency for CO₂ dissociation (up to 40–50%) [9,10,11,12,13,14,15,16], which makes them attractive from an industrial perspective.

The classic GAD consists of two diverging (knife-shaped) electrodes and a gas nozzle. It is a non-stationary discharge with a cyclic evolution. The arc ignites at the closest electrode distance and is then pushed downstream by the gas flow and “glides” along the electrodes towards the region with a larger interelectrode distance. As a result, its length increases; when a critical length is reached, the plasma channel extinguishes, and a new arc is ignited at the closest electrode distance. The repetition rate of the cycle and the length of the discharge depend on the gas flow, the current and the magnetic field, if present.

In terms of CO₂ conversion, only a limited amount of the gas is usually processed (~20%) [12]. To overcome this limitation, different discharge constructions have been investigated. The rotating gliding arc with inner and outer cylindrical electrodes forming a tornado-like discharge [17] and the gliding arc plasmatron with a reverse vortex flow [18,19,20] have been developed.

Several attempts have been made to improve the performance of the classic GAD through various modifications. One of the modifications studied is the implementation of a magnetic field perpendicular to the arc current [15,18,21,22,23,24]. The magnetic field induces a J × B force that acts either upstream or downstream of the gas flow, depending on the current and the magnetic field directions.

This work is a continuation of a previous investigation [21] studying three different discharge configurations: classic gliding discharge (GD); magnetically accelerated gliding discharge (MAGD); and magnetically stabilized gliding discharge (MSGD). The classic GD uses the standard design with flat diverging electrodes, as described above. The MAGD is a modification of the GD with an external magnetic field in a direction perpendicular to the arc current accelerating the arc downstream. The MSGD utilizes a magnetic field that generates J × B force opposite to the gas flow, stabilizing the arc at a specific position on the vertical axis. The focus of the research was mainly on the MSGD and its potential for CO₂ conversion. The physics of magnetic stabilization were also previously examined by numerical modelling for the arc plasma [23]. The experimental results in Ivanov et al. [21] for CO₂ conversion showed a good conversion rate and energy efficiency for MSGD at low gas flow rates (for 2 L/min—conversion rate about 8%); however, these values decrease rapidly as the flow rate increases. The classic GD configuration studied at the same gas flow rates (2–12 L/min) and discharge currents (50–200 mA) shows better results at high gas flow rates compared to MSGD and MAGD.

In this study the discharge configurations were upgraded with an active cooling system, which is used to improve the cooling rate of the arc and the gas in the afterglow. In all configurations (see Figure 1), the arc is placed between two quartz glasses to channel and confine the gas flow and the arc motion. To improve the CO₂ conversion, the cooling rate is modified by changing the distance d between the plates and the electrode thickness, thus changing the heat transport of the gas to the actively cooled glasses. The variation of the channel thickness affects the gas velocity for a given gas flow rate and, respectively, the gas temperatures (T_g). It has been shown that reducing T_g lowers the reverse reaction (CO + O₂ → CO₂ + O) and enhances the CO₂ conversion rate and energy efficiency [13]. This effect is also confirmed in [25,26] by rapid gas quenching of the treated gas. Of course, it should be taken into account that cooling the plasma channel region may reduce the gas temperature there and reduce the forward reaction of CO₂ dissociation.

The discharges studied here have two important advantages, as follows: (1) the use of a magnetic field allows for the control of the arc motion and the relative velocity between the arc channel and the surrounding gas, which, as shown here, considerably improves the CO₂ conversion; and (2) the use of bounding dielectric walls allows for further control and optimization of the gas (arc channel) cooling.

In addition to the channel thickness study, the gas flow and the magnetic field distribution are commented on, and the results are supported by 3D numerical modelling of the gas flow. Note that the previous research [21] and studies in the literature [27] show that at low currents of a few hundred mA, the discharges can operate in two different regimes with respect to the cathode region electron emission process—the arc and contracted glow discharge. Therefore, the term “gliding arc discharge” will be avoided; instead, the more appropriate and general term “gliding discharge (GD)” will be used. Throughout the text, the term “arc” is often used, implicitly referring to both regimes. In general, glow and arc discharges could also be distinguished based on the behavior of the current–voltage (I–V) characteristics (flat or falling, respectively); however, the cathode emission is considered as the defining property in this study.

2. Materials and Methods

2.1. Experimental Setup

A schematic of the experimental setup is shown in Figure 1a. The system operates at atmospheric pressure (around 700 Torr in Sofia, Bulgaria), as the plasma reactor is connected to the atmosphere. The discharge device is situated in a glass vacuum-sealed tube. The gas flow is introduced to the discharge device via a system of pipes, and it is controlled by a mass flow controller Bronkhorst EL-FLOW F-201CM. The gas flow rate is measured in normal L/min (temperature 273.15 K and pressure 760 Torr). Part of the outflow gas is sampled by a pump to a gas analysis system, which consists of a Fourier spectrometer (FT-IR) Perkin Elmer Frontier operating in the range of 1000–6000 cm⁻¹ equipped with 10 cm gas cell Specac Storm 10. The absorption line, 2209 cm⁻¹ (4527 nm) of the CO molecule band, is used to determine the conversion rate of CO₂ into CO and O₂.

The discharge device is powered by a high-voltage (HV) power supply consisting of 3 transformers (SIET Metalbox 10 kV), a full bridge rectifier, and an LR filter with L = 20 H inductance (Figure 1a). The transformers have an internal limitation of the maximum current based on a magnetic shunt in their magnetic core. The maximum current values are 50, 100, and 210 mA. A more detailed scheme of the power circuit can be found in Ivanov et al. [21]. The power supply is not a perfect constant voltage/current source, and it has a considerable oscillating component at 100 Hz, despite the large inductive filter installed. As a result, the spectrum of the voltage signals is rather complicated, with several frequency components, including 100 Hz and other components determined by the gas velocity and the arc cycle. The voltage and current modulation, at 100 Hz, is beneficial in the magnetically retarded GD (MRGD) configuration since it avoids firm stabilization of the arc at a certain “z” position due to the variation of the current and therefore the variation of the J × B force, causing the arc to oscillate along “z” (a more detailed explanation of the discharge configurations can be found in Section 2.2). The latter avoids the overheating of the dielectric (quartz) windows when the arc is stabilized.

The current is measured with a clamp-on current probe (Pintek PA-699) and the voltage is measured with a differential HV probe (Pintek DP-30 K), both connected to an oscilloscope (Rohde & Schwarz^®RTB2004, Praha, Czech Republic, Digital Oscilloscope). A PC is used to calculate input power, conversion rate and energy efficiency.

An example of the obtained discharge current and voltage is shown in Figure 2.

The discharge device, shown in Figure 1b, uses a design based on the classic GD with flat, solid, knife-shaped electrodes placed between two dielectric quartz walls with a thickness of 2 mm. The electrodes are made of copper and have a thickness equal to the channel thickness, i.e., for every channel thickness value, a different pair of electrodes is used. A system for water cooling is placed over the quartz glass plates. It consists of 4 mm-thick aluminum plates and copper tubes with flowing water. The copper tubes cool the aluminum plates with sufficient thermal contact. For better heat transfer, a thermal paste is applied between the electrodes and the quartz glass walls as well as between the quartz and the aluminum plates of the cooling system. Permanent neodymium magnets (B_max = 0.2 T) are placed over the aluminum plates of the cooling system in the configurations with a magnetic field. The magnetic field distribution is presented in Section 2.2. The external magnetic field is used to control the arc and the operating regime of the discharge. The active water-cooling system is used to improve performance and allow for the long-time operation of the discharge. The magnets are also cooled down to avoid temperature increases above the Curie point and the loss of their magnetic properties.

2.2. Discharge Configurations

Three different discharge configurations are being examined: classic GD, MAGD, and MRGD. The difference between the last two is due to the orientation of the magnetic field. In the case of MAGD, the external magnetic field is in a direction that accelerates the arc downstream due to J × B drift. In the case of MRGD, the magnetic field is oriented in a way that retards the gliding of the arc, but it is not strong enough to stabilize it.

GD is the classic configuration of a gliding discharge between two knife-shaped electrodes without an external magnetic field. The discharge and the gas flow are additionally channeled by two quartz glasses, as shown in Figure 1b. The gas flow inlet is located at the bottom of the discharge device. The arc is ignited at the shortest distance between the electrodes, which is 3.5 mm.

Two different sets of magnets, and therefore, the magnetic field distributions, are used to analyze the effect of the magnetic field, as follows: (1) magnets 50 × 5 × 10 mm (xyz) as shown in Figure 1b, producing a magnetic field distribution, as shown in Figure 3a and hereafter referred to as MFD1; and magnets 40 × 7.4 × 60 mm (not shown in Figure 1b), referred to as MFD2 (Figure 3b). The difference is mainly in the spatial extent of the magnetic field—MFD1 is localized in a narrow region, while MFD2 affects the arc motion in the whole domain. The 3D FEM simulation parameters are adjusted in order to provide experimentally measured values of the magnetic field at the center of a discharge device with an electrode thickness of 1 mm. When the channel thickness is increased up to 4 mm, the magnetic field is reduced by up to 20% for the MFD1 and by up to 8% for the MFD2.

2.3. Quantities of Interest

The main quantities studied are the conversion rate and the energy efficiency, both as a function of the flow rate and the specific energy input (SEI). The conversion rate is defined as the fraction of the converted CO₂ into CO, as follows:

$$Conv.rate\left[\mathrm{\%}\right]={\displaystyle \frac{n_{C{O}_2}^{in}-n_{C{O}_2}^f}{n_{C{O}_2}^{in}}}\times 100\mathrm{\%},$$

(1)

where $n_{C{O}_2}^{in}$ and $n_{C{O}_2}^f$ are the initial and the final concentrations of CO₂, respectively. This fraction is obtained through the relative absorbance of a sample of the processed gas at a line of the CO molecule, at 2209 cm⁻¹. The intensity of the line is proportional to the concentration of CO in the mixture. To determine the absolute conversion rate, the intensity of the line is multiplied by a coefficient, which is found through separate measurements using a calibration gas mixture containing 20% CO and 80% CO₂. Due to the splitting of CO₂ into CO and O₂, the volume increases by a factor of 1.5. In order to account for this expansion, a correction is applied. Further details can be found in Ivanov et al. [21].

Another parameter used is the SEI, which is the energy delivered per mole of gas, as follows:

$$SEI\left[{\displaystyle \frac{\mathrm{J}}{\mathrm{m}\mathrm{o}\mathrm{l}}}\right]={\displaystyle \frac{P\left[\mathrm{J}/\mathrm{s}\right]}{MFR[\mathrm{L}/\mathrm{s}]\times \left(1/22.4\right)\left[\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{L}\right]}},$$

(2)

where $MFR$ is the mass flow rate of the gas.

The energy efficiency is the fraction of energy required for the conversion of CO₂ to the total energy input ($SEI$), as follows:

$$\eta \left[\mathrm{\%}\right]={\displaystyle \frac{Conv.rate\times ∆H_R}{SEI}}\times 100\mathrm{\%},$$

(3)

where $∆H_R=279.8\times {10}^3[\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l}]$ is the reaction enthalpy for the CO₂ splitting reaction. The absolute uncertainties for the conversion rate are calculated using the formula from Ivanov et al. [21], as follows:

$${\left(∆X_{C{O}_2}\right)}^2={\displaystyle \frac{36}{{\left(3-X_{C{O}_2}^0\right)}^4}}{\left(∆X_{C{O}_2}^0\right)}^2,$$

(4)

where $X_{C{O}_2}^0$ is the conversion rate before correction and the $X_{C{O}_2}$ is the real conversion rate. The relative uncertainty in the energy efficiency is calculated as follows:

$$\left|∆\mathsf{\eta }/\mathsf{\eta }\right|=\left|∆X_{C{O}_2}/X_{C{O}_2}\right|+\left|∆SEI/SEI\right|,$$

(5)

where $\left|∆SEI/SEI\right|\approx \left|∆MFR/MFR\right|+\left|∆P/P\right|$ and $P$ is the discharge power. More details of the procedure for calculating the uncertainty of the results obtained can be found in Ivanov et al. [21]. The relative uncertainty for the CO₂ conversion rate is in the range of 2–3%; the relative uncertainty for the energy efficiency is about 12–13% of the energy efficiency value. Note that, while the uncertainty of the efficiency is relatively high, our studies show that the repeatability, and thus the precision, remains much smaller, enabling us to make conclusions about the relative changes in the trends of the efficiency at different conditions. Note also that the uncertainty is not shown at all points in some of the following figures in order to avoid overloading and difficulty in reading the figures.

2.4. Gas Flow Modelling

One of the key parameters studied in this work is the channel thickness, which significantly changes the gas flow velocity. For a better understanding of the experimental results obtained, it is useful to know the gas velocity distribution and magnitude. Here, the results from 3D simulations of the gas flow, without any gas discharge, are added and presented in Figure 4 and Figure 5. Note that the discharge can modify the gas flow distribution; therefore, these modelling results are approximate and are used only as a rough estimation of the gas velocity variation in different cases with different channel thicknesses. The gas velocity profile can be modified, for example, in the case of a retarded arc, due to the additional drag caused by the arc pushed upstream by the J × B force. Moreover, the gas heating due to the plasma column leads to gas expansion and gas acceleration in the post-discharge zone, i.e., the gas in the downstream region should move faster compared to the behavior presented in Figure 5.

The results are obtained using the numerical package COMSOL Multiphysics version 6.2 with the SST RANS (shear stress transport Reynolds-Averaged Navier–Stokes) turbulence model [28,29,30] in CO₂ gas. A turbulent model was used due to the large variation of the Reynolds number from 130 to 7000 for the different configurations and conditions. The SST model describes the gas flow in the whole domain, including the boundary layers near the walls. The boundary conditions are as follows: at the walls “no slip” (zero tangential velocity); at the outlet the pressure is fixed to atmospheric; and at the gas inlet, a parabolic velocity profile is imposed, with the peak velocity corresponding to the experimental gas flow.

Figure 4 shows the 2D distribution of the gas flow in the XZ plane at y = 0, while Figure 5 shows the gas velocity magnitude along the Z axis, i.e., in the middle of the domain.

The distribution shown in Figure 4 is similar, qualitatively, for the different channel thicknesses and gas flow rates. However, the scaling of the velocity is not exactly proportional for different channel thicknesses, as shown in Figure 5. This is due to the effects related to the rounding of the electrodes, which changes the effective cross-section, and to the different turbulent energy, due to the different Reynolds number.

It is worth noting the very inhomogeneous gas velocity distribution in the upper part of the discharge observed in Figure 4—the gas flow is concentrated in a narrow channel and the arc is subject to a different velocity along its length.

3. Results and Discussion

This section presents results for the influence of different external parameters, such as the arc current, magnetic field distribution, and channel thickness on the CO₂ conversion and energy efficiency. The parameters and operating conditions are summarized in

Table 1. Main parameters and their range/values.

Parameter	Range/Values
Gas flow rate	2–12 L/min
Discharge currents	50, 100, 210 mA
Channel thicknesses	1, 2, 3, 4 mm
SEI	0.35–14 kJ/L
Average power	80–650 W
Average voltage	1.8–5 kV

. The role of channel thickness and the presence of permanent magnets are investigated.

The results for the three configurations of the discharge (GD, MAGD and MRGD) are presented and analyzed, focusing on the effect of the magnetic field on the difference between the arc and the gas velocity, and its impact on the gas treatment.

3.1. The Influence of the Channel Thickness

The channel thickness mentioned above refers to the distance between the dielectric flat plates and also to the thickness of the electrodes. These plates, and the electrodes, form a sandwich-like structure. Their main function is to confine the gas flow; however, the distance between them also affects the gas flow velocity and, consequently, the heat transfer between the gas and the quartz walls.

Figure 6 and Figure 7 show the energy efficiency as a function of the conversion rate for arc currents 100 and 210 mA, respectively. They show the trends for the three discharge configurations using the localized magnetic field distribution MFD1 for the studied range of gas flow rates and channel thicknesses. The results are at different mass flow rates, but the flow rate itself is not shown in the figures. In the following subsection, the magnetic field distributions will be differentiated and compared. The results for the arc current of 50 mA are not presented here, as they show lower conversion rates at similar efficiencies and similar trends and, thus, do not provide any advantage.

The values determined for the conversion rate and energy efficiency are in the range of 2–8% and 3–37%, respectively. The combination of the sufficient conversion rate and high energy efficiency is the desired operating condition for all discharges used to convert CO₂. Although there are no strict criteria for “the best” results, here, we consider the relative joint maximization of both quantities as the most promising results of interest.

With respect to the channel thickness, Figure 6 and Figure 7 show that the 3 mm case (red) has overall better combinations of the two parameters—conversion rate and energy efficiency—compared to those of 1, 2 and 4 mm. The 1 mm case (blue symbols) shows the worst performance of all cases. It is supposed that, at a small distance of 1 mm, the walls are intensively cooling the arc channel, causing further arc contraction; thus, the region of sufficiently high temperature for CO₂ dissociation is shrunk compared to the larger distances. In addition, the proximity of the walls generally leads to additional losses of charged and excited particles, further reducing the dissociation rate. With the increase in channel thickness, the results tend to improve for all configurations, with the best results obtained at 3 mm. A further increase to a 4 mm-channel thickness (purple symbols) leads to a small deterioration in performance. In this case, the gas channel becomes significantly wider compared to the visible arc diameter (about 1–1.5 mm); therefore, a smaller amount of the flowing gas is treated. This reduces the conversion rate, while the energy efficiency remains similar to the 2 mm and 3 mm cases. In Figure 7c, for the MRGD case, there are no results for the 4 mm-channel thickness because the arc stabilizes at these conditions.

For the sake of completeness, the results for the conversion rate and the energy efficiency as a function of the mass flow rate for the discharge configuration MAGD–MFD1 at varying channel thicknesses are shown in Figure 8. The results confirm the well-known trend for GAD [15,21,22], that the conversion rate decreases with the gas flow rate and the energy efficiency increases with the gas flow rate.

However, as discussed in Section 2.4, the gas flow profile in the classic GD design is highly inhomogeneous, which results in varying conditions along the arc length and complicates the optimization process. Improving performance in certain regions may reduce it in others. A potential solution could be a modified design that promotes a more homogeneous gas treatment along the arc length, which might enhance the overall performance.

With respect to the effect of the arc current value, Figure 9 shows that, when plotted as a function of the SEI, the results are similar to those typically reported in the literature [4,21]: the conversion rate increases with the SEI, i.e., with the arc current and thus with the power, and the energy efficiency decreases at a higher SEI (arc current and power). The other two configurations (GD and MRGD) show similar trends and thus are not shown here.

In the following subsections, only results for the cases of d = 2 and 3 mm will be presented in order to keep the figures easily readable; moreover, these cases show the best overall performance.

3.2. The Influence of the Magnetic Field

3.2.1. Comparison of GD, MAGD, MRGD Under the Same Conditions

Here, the effect of the magnetic field in configuration MFD2 on the energy efficiency and the conversion rate is presented (Figure 10). For the same SEI, the conversion (Figure 10a) is almost doubled at SEI values of 10–12 kJ/L and about 50% at the lower SEI values. The configurations with a magnetic field lead to the relative motion of the arc column with respect to the gas flow and thus tend to treat more gas. On the other hand, both configurations with magnetic fields show very similar performances, which again indicates the importance of the relative velocity between the arc and the gas compared to the absolute velocity of the arc with respect to the walls. This increase in the conversion rate is accompanied by an increase in the energy efficiency in the order of 30% (relative) at a high SEI and 10–15% (relative) at a lower SEI. It is expected that the faster quenching in the magnetic configurations is contributing to the increase in the overall efficiency as it enhances the quenching, probably by creating a more turbulent afterglow, which increases the turbulent thermal conductivity.

3.2.2. Influence of the Magnetic Field Distribution

To further investigate the effect of the relative velocity, both magnetic field distributions are examined. In MFD1, the magnetic field is localized in a narrow region, whereas in MFD2, it is significant over a large region (see Figure 3). This results in different behaviors of the arc in the two configurations and in different rates of performance (see Figure 11). If the MAGD configuration (accelerated arc) is considered, in Figure 11a, one can see a significant difference in the obtained results—the performance of the MFD1 (localized MF) is worse than that of MFD2 and is similar to that of the GD configuration without a magnetic field. It is supposed that the reason for this behavior is related to the average time spent with significant relative velocity. In the MFD2 configuration (large magnetic field region), the arc column is pushed downstream in the whole domain (arc path), which ensures significant relative velocity between the arc and the gas during the whole arc column path. In the MFD1 case (localized field), the arc operates with significant relative velocity for a short time and only in a small region, which means that the arc will operate as a GD without a magnetic field for a significant amount of time. Therefore, the localized magnetic field configuration MFD1 shows a performance similar to that of the GD without magnetic field.

On the other hand, the retarded configuration (MRGD) in both magnetic field configurations ensures significant time with significant relative velocity. Therefore, the results presented in Figure 11b are rather similar for both magnetic distributions and deviate only in the case of low gas flow rates (the green points with the lowest efficiency). Even in the case of a localized magnetic field (MFD1), the time spent in the magnetic field region with high relative velocity remains significant compared to the time spent in the no-magnetic field regions with low relative velocity.

The magnetic field and the increased relative velocity arc–gas have a positive effect on the CO₂ treatment. One could further enhance this effect by working at higher currents and gas flow rates and thus at higher relative velocities. We were not able to verify this due to technical limitations.

As mentioned above, the discharge cycle of ignition, extension, decay, and re-ignition has a significant impact on the discharge behavior and CO₂ conversion. This cycle is strongly affected by the gas flow velocity, but also by the power supply. As mentioned in Section 2, the power supply has a considerable oscillating component at 100 Hz. In general, this is advantageous since it makes the arc stabilization at a certain point more difficult due to the oscillating Lorentz force and thus avoiding melting of the dielectric walls; however, it also influences the arc cycle of ignition and decay, which is no longer related to the gas flow velocity. When the gas velocity and the discharge expansion velocity are sufficiently high, the discharge expansion cycle is much shorter than 10 ms. In this case, additional frequency components start to appear in the voltage spectrum, i.e., for a single period of the power supply signal, several periods of the discharge cycle take place. For example, for a channel thickness of 2 mm and a gas flow rate of 6 L/min, the gas velocity in the discharge expansion region is approximately 14 m/s (see Figure 5), which leads to 2–3 discharge cycles per 10 ms for the GD configuration, approximately 1.5 cycles per 10 ms in the case of MRGD–MFD1, and around 3.5 discharge cycles in the case of MAGD–MFD1. In the case of retarded arc MRGD with MFD2 (wider magnetic field region), the arc travels all the time in a significant magnetic field and the cycle frequency is determined by the power supply (100 Hz). The MAGD–MFD2 configuration at 6 L/min, 100 mA oscillates mainly at the power supply frequency, i.e., one discharge cycle per 10 ms. This is surprising when compared to the 3.5 cycles observed in the MAGD–MFD1 configuration; the reason is the tendency of the MAGD–MFD1 to be subject to back-breakdown, i.e., breakdown between the rising branches of the arc column that are sufficiently close to each other and shortened. Thus, as the discharge is not reignited at the shortest electrode distance, the expansion path, and, therefore, the time is reduced, leading to the observed higher frequency.

In the case of a channel thickness of 3 mm and a gas flow rate of 6 L/min, the gas velocity is reduced; the discharge cycle period is very close to 10 ms. As a result, the discharge motion is determined mainly by the power supply frequency.

4. Conclusions

This work presents systematic research on the performance of the following three different discharge configurations: classic gliding discharge with diverging electrodes; magnetically accelerated gliding discharge; and magnetically retarded gliding discharge. All these use dielectric walls to confine the gas flow and the arc operation in a narrow channel between the electrodes. The effects of wall cooling and channel thickness, as well as the magnetic field distribution, were examined.

The results at different channel thicknesses show the following: (1) the channel thickness strongly affects the conversion rate and the energy efficiency. A small distance of 1 mm is detrimental to the overall performance and shows the worst results, which is expected to be due to the lower average gas temperature and the additional heat and charged particle losses at the walls; (2) channel thicknesses of 2 and 3 mm show the best overall performance for the discharge configurations considered; and (3) a channel thickness of 4 mm leads to a slightly reduced conversion because the gas channel becomes much larger compared to the arc diameter.

The discharge configuration also has a significant effect on the performance. The use of a magnetic field to accelerate or delay the arc channel with respect to the surrounding gas flow (MAGD and MRGD, respectively) has a significant effect compared to GD without a magnetic field. At certain conditions, this results in a 40–50% relative increase in the conversion and a 30% relative increase in the energy efficiency. The comparison between different magnetic field distributions shows the importance and benefit of the relative velocity between the arc channel and the surrounding gas for the observed improvement. Moreover, the use of active cooling leads to better stability of the discharge compared to that of previous studies.

In comparison to other research on gliding arc discharges, the results obtained here show similar or improved performance. For example, in Sun et al. [12], a 6–8% conversion rate at 25–35% energy efficiency was reported, which is very close to the results obtained here. It is also worth noting that this discharge achieves results on par with the gliding arc plasmatron with reverse vortex gas flow stabilization reported in Ramakers et al. [20].

Overall, this study shows that the classical gliding arc discharge can be further improved by simple modifications with the use of a magnetic field, which adds another degree of freedom, providing additional possibilities for optimizations and improvements. The latter is attributed mainly to the significant relative velocity between the arc channel and the gas on the CO₂ conversion.