Investigation of flow characteristics in a twin-surface dielectric barrier discharge reactor by Schlieren imaging

The SDBD electrode configuration consisted of an aluminum oxide (α-Al₂O₃) plate with two identical Ni-plated metal grids screen-printed on each side. The α-Al₂O₃ plate had dimensions of 190 mm × 88 mm × 0.635 mm and acted as the dielectric barrier. The metal grid lines had a width of 0.45 mm, a height of approximately 30 µm and formed evenly spaced squares of 10 mm × 10 mm. The metal grids consisted of 15 × 5 squares and served as the powered and grounded electrode of the SDBD. A photograph of the SDBD electrode configuration is shown in figure S1 (supporting information). A commercial high-voltage (HV) pulse generator with an external transformer (Redline G2000, Redline Technologies, Germany) provided HV pulses with typical peak-to-peak values ${U_{{\text{pp}}}}$ between 7 kV and 11 kV and a defined pulse repetition frequency ${f_{{\text{pul}}}}$ of 4000 Hz to the powered electrode. The resulting voltage waveform had the shape of a pulsed damped sine wave. The natural frequency of 86 kHz was defined by the secondary inductance of the HV transformer and the capacitance of the SDBD electrode configuration, which formed a series resonant circuit. The plasma ignited along the grid lines on both sides of the dielectric surface.

All measurements were performed at ambient pressure and temperature in an all stainless-steel flow setup with a custom-made reactor. Three different gas lines with H₂ (purity > 99.999%, Air Liquide), N₂ (purity > 99.999%, Air Liquide) and O₂ (purity > 99.998%, Air Liquide) were connected to the flow setup. The gas flows were adjusted by calibrated mass flow controllers (EL-FLOW Select, Bronkhorst High-Tech B.V., Netherlands) and were mixed to produce the desired gas mixtures. A pneumatically actuated four-port-two-position valve (Valco valve, VICI AG International) was used to set the complete reactor unit either online or bypass. The reaction chamber was presented by Peters et al [29] in detail and was made from aluminum. KF 25 flanges at its inlet and outlet expand to a rectangular cross section of 105 mm × 17 mm with a length of 200 mm. The SDBD electrode configuration was positioned at the center of this rectangular reaction chamber and insulated from the reactor walls by PEEK mountings. Therefore, the gas flow passed along both sides of the SDBD electrode configuration. Rectangular quartz windows (Viosil SQ, GVB GmbH, Germany) with dimensions of 50 mm × 20 mm were placed on the center of the flat sides of the reaction chamber (17 mm × 200 mm) and allowed a side view of the electrode configuration. The complete reaction chamber was grounded. Voltage and current measurements were performed with a HV probe (P6015A, Tektronix, USA) and a current probe (Model 6585, Pearson Electronics, USA) connected to a digital oscilloscope (DPO5204B, Tektronix, USA). A typical current and voltage measurement is shown in figure S4. The composition of the exhaust gas stream was monitored online by a multi-channel analyzer (MCA) (X-STREAM Enhanced XEGP, Emerson, USA) equipped with an electrochemical O₂ trace sensor and a thermal conductivity detector for H₂. Due to the plasma discharge a minor temperature increase of the bulk gas to approximately 50 °C is observed. More detailed temperature measurements are described by Peters et al [29]. A schematic illustration of the reactor unit is shown in figure 1. Photos of the reactor with an active plasma discharge are shown in figure S5.

Zoom In Zoom Out Reset image size

If not stated otherwise, a mixture of 500 ppm O₂ balanced in N₂ with a volumetric flow rate of 10 l min⁻¹ was used resulting in a linear flow velocity of approx. 0.1 m s⁻¹ in the rectangular reaction chamber. For the reaction of O₂ traces with H₂ to H₂O, a mixture of 500 ppm O₂ balanced in a H₂/N₂-mixture with an identical flow rate was used. The degree of conversion was calculated as the ratio of the converted amount of O₂ relative to its amount in the feed. The HV pulses were applied to the top electrode. The addition of toluene was carried out using a controlled evaporator mixer (CEM, Bronkhorst High-Tech B.V., Netherlands) system, through which the respective N₂ flow was directed.

The current–voltage characteristics of the SDBD have already been described by Offerhaus et al [26] and Kogelheide et al [30]. Plasma ignition can be observed as numerous spikes on the first half-waves of the current waveform. The power dissipated into the system was derived as described by Schücke et al [27] (equation (1)),

$$\begin{align}{P_{{\text{diss}}}} = {f_{{\text{pul}}}}\int\limits_{{T_{{\text{pul}}}}}^{} {u_{{\text{meas}}}}\left( {{i_{{\text{meas}}}} - {C_{{\text{sys}}}}\frac{{\partial {u_{{\text{meas}}}}}}{{\partial t}}} \right){\text{d}}t.\end{align} \tag{ 1 }$$

Here, ${u_{{\text{meas}}}}$ is the measured voltage, ${i_{{\text{meas}}}}$ the measured current, ${T_{{\text{pul}}}}$ the duration of a pulse period and ${C_{{\text{sys}}}}$ the total capacitance of the system. It has to be noted that the power dissipated in the system is not equivalent to the power dissipated in the plasma, but additionally includes ohmic losses in the conductors and dielectric losses in the dielectric plate of the electrode configuration.

A spray coating procedure previously presented by Peters et al [29] was used to create a thin and homogeneous film of platinum nanoparticles (<50 nm, Sigma Aldrich) on both sides of the electrode configuration. The desired loading of 150 µg cm⁻² was obtained using the following procedure: a suspension of 0.05 mg ml⁻¹ Pt nanoparticles in a 1:1 mixture of isopropanol and water was prepared and treated for 40 min in an ultra-sonic bath. This suspension was sprayed onto the electrode configuration with a volume of 30 µl per spray point and 1 mm between the spray points. The liquid was immediately evaporated as the target was heated to 200 °C.

Schlieren imaging was used to visualize density gradients of the gas phase inside the reaction chamber. Schlieren images were recorded by a lens-type Schlieren setup similar to the instrument described by Settles [25] and adapted for the SDBD by Offerhaus et al [26]. A schematic illustration of the Schlieren setup is shown in figure 2. A point-like light source was created by focusing a monochromatic LED light source onto a pinhole using a condenser lens ($\;f\, = \,50\,{\text{mm}}$). Another lens ($\;f\, = \,200\,{\text{mm}}$) was used to collimate the light, and the SDBD reactor was placed within the parallel light beam. The light beam passed through the side windows of the reaction chamber and was therefore perpendicular to the flow direction. Consequently, several grid lines were in line of sight. After the light beam had passed through the SDBD reactor, another lens ($\;f\, = \,200\,{\text{mm}}$) focused it onto a razor blade (Wilkinson, UK), which was used as the knife edge of the Schlieren aperture. The knife edge was oriented in such a way that vertical Schlieren images were obtained. All images were recorded by a commercially available camera (Canon EOS 6D) equipped with a zoom lens (Canon 24–105 mm f/4.0) and a resolution of 20 MP. For each voltage step the images were recorded with an exposure time of 5 ms after 5 min of operation.

Zoom In Zoom Out Reset image size

The recorded images were processed by using a MATLAB (MathWorks, Massachusetts, USA) script. The RGB (red, green, blue) images were converted into 256 bit grayscale versions. The resulting images were subtracted from a background image, which was recorded without plasma ignition. To increase the contrast, the final image was rescaled to the full 256 bit grayscale. The individual steps of the data processing are shown in figure S2.

The twin SDBD modeled in this work is based on the experimental setup described in section 2. For simplicity, we consider only the plasma chamber where the plasma discharge phenomenon is observed. The plasma chamber was defined as the computational domain for fluid flow and the dielectric medium positioned at the center as a solid domain to capture the plasma discharge effects. Due to the large difference in the length scales between the height of the electrodes and the length of the plasma chamber in the computational domain, 3D modeling is not viable and hence a 2D model along the flow direction was developed. The 2D model consists of a solid domain representing the dielectric medium and a fluid domain for the gas flow region as shown in figure 3.

Zoom In Zoom Out Reset image size

An unstructured mesh was generated for the computational domain with enhanced meshing in both the solid domain to capture the electric field strengths between the electrodes, and the fluid domain near the electrodes to capture the flow details close to the electrodes. A total number of 342 410 mesh elements with a minimum mesh size of 2 μm were generated. Orthogonal quality was the mesh metric used to determine its quality. A value of 0.28 which lies in the 'good' range and above the 'acceptable' range in the ANSYS orthogonal quality mesh metrics spectrum was achieved [31].

The simulation of gas flow inside the 2D model of the plasma chamber was carried out with N₂ gas with an inlet velocity of 0.1 m s⁻¹ corresponding to a flow rate of 10 l min⁻¹ in the experimental setup. In order to capture the effects of convective heat transfer, it is assumed that density is a function of temperature only. Since the flow rate corresponds to the low Reynolds number flow regime, a laminar flow model was used, which solves the Navier–Stokes equations to obtain the velocity profile. The inlet velocity and atmospheric pressure were set as the inlet boundary condition and outflow boundary conditions at the outlet. The electrodes, dielectric surface and the walls are subjected to no-slip boundary condition. The governing equations for the fluid flow are described below.

Continuity equation

$$\begin{align} {\frac{{\partial {{\rho }}}}{{\partial {{t}}}} + \nabla \cdot\left( {{{\rho u}}} \right) = 0\,} .\end{align} \tag{ 2 }$$

Momentum equations

$$\begin{align} {\frac{{\partial \left( {\rho u} \right)}}{{\partial t}} + \nabla \cdot\left( {\rho u{\boldsymbol{{u}}}} \right) = - \frac{{\partial p}}{{\partial x}} + \nabla \cdot\left( {\mu \nabla u} \right) + \rho {g_x} + {F_{x,{\text{ ext}}}}{ }} \end{align} \tag{ 3 }$$

$$\begin{align} {\frac{{\partial \left( {\rho v} \right)}}{{\partial t}} + \nabla \cdot\left( {\rho v{\boldsymbol{{u}}}} \right) = - \frac{{\partial p}}{{\partial y}} + \nabla \cdot\left( {\mu \nabla v} \right) + \rho {g_y} + {F_{y,{\text{ext}}}}{ }} \end{align} \tag{ 4 }$$

$$\begin{align} {\frac{{\partial \left( {\rho w} \right)}}{{\partial t}} + \nabla \cdot\left( {\rho w{\boldsymbol{{u}}}} \right) = - \frac{{\partial p}}{{\partial z}} + \nabla \cdot\left( {\mu \nabla w} \right) + \rho {g_z} + {F_{z,{\text{ext}}}}{ }} .\end{align} \tag{ 5 }$$

3.3.1. Governing equations.

The plasma discharge model for a twin SDBD is based on the simulation approach of plasma actuators used in aerodynamic applications [32]. A phenomenological body force model developed by Suzen et al was often used to study the plasma discharge effects on fluid flow in plasma actuators [33, 34]. In this model, it is assumed that the charges in the plasma have sufficient time to redistribute and the whole system is quasi-steady with no magnetic field effects. It is also assumed that the potential is decomposed into two parts: the potential due to the external field and the potential due to the net charge density. This results in the following two Maxwell equations, which solve for electric potential and charge density,

$$\begin{align} {\nabla \cdot\left( {\varepsilon \nabla \varphi } \right) = 0\,} \end{align} \tag{ 6 }$$

$$\begin{align} {\nabla \cdot \left( {{\varepsilon _{\text{r}}}\nabla {\rho _{\text{c}}}} \right) = - \frac{{{\rho _c}}}{{\lambda _{\text{d}}^2}}{} } .\end{align} \tag{ 7 }$$

Furthermore, the electrohydrodynamic body force generated due to the plasma discharge is expressed in terms of charge density and electric field strength which are then coupled with the Navier–Stokes equations as an external body force acting on the bulk fluid [33, 34]. Experimentally, the plasma was ignited by a pulsed voltage input with a pulse repetition frequency of 4 kHz and voltage pulse amplitude of 7 kV–11 kV, which results in the generation of a pulsed body force. For the purpose of modeling and to investigate the plasma discharge effects on fluid flow, only the voltage amplitude of 10 kV was used. Jayaraman proposed a time-averaged body force model to calculate the instantaneous body force developed due to the pulsed discharge and is given in equation (8) [35],

$$\begin{align} {{F_{{\text{ext}}}} = \frac{{\Delta t}}{{{t_{\text{p}}}}}{\rho _{\text{c}}}E{} } .\end{align} \tag{ 8 }$$

The transfer of thermal energy in a plasma discharge resulting in gas heating occurs through multiple processes. The energy transfer occurs between electron and neutral species by elastic collisions, rotational, vibrational, and electronic excitation and also through collision between charged particles and neutral species, which result in gas heating at different timescales [36, 37]. Due to the complexity of the gas heating mechanism and to additionally incorporate the energy transfer into the plasma discharge model, the experimental data of power deposited during the plasma discharge was considered. In this approach, it was assumed that the power deposited during the plasma discharge process is utilized only in heating of the bulk fluid by means of heat flux from the electrodes. Here, the power deposited was assumed to be 45 W and is given as the source term on the electrodes. Thus, the energy equation is solved with an external heat source term,

$$\begin{align} {\frac{{\partial \left( {\rho {C_{\text{p}}}T} \right)}}{{\partial t}} + \nabla \cdot\left( {\rho {C_{\text{p}}}{\boldsymbol{{u}}}T} \right) = \nabla \cdot\left( {k\nabla T} \right) + Q{ }{\text{}}}.\end{align} \tag{ 9 }$$

3.3.2. Numerical implementation.

In order to study the gas heating effects on the fluid flow, a gas flow model with heat transfer mechanism through heat flux from the electrodes and without the body force model was developed. In this model, the Navier–Stokes equation and the energy equation with external heat source was solved simultaneously. The same boundary condition used in the model mentioned above was applied.

Previous investigations by Schücke et al [27, 28] and Peters et al [29] demonstrated that high degrees of conversion can be obtained with the presented SDBD reactor applied in the total oxidation of VOCs in synthetic air. As the plasma is limited to the surface of the SDBD electrode configuration (100 µm plasma height) and has a lateral extension of only a few millimeters, while the plasma-free volume is substantially larger, these high degrees of conversion must be due to significant radial mass transport. For this study the reaction of O₂ traces with H₂ was chosen as fast and thermodynamically highly favorable reaction with only a single product. Figure 4(a) shows the O₂ mole fraction as a function of time in a gas mixture of 100% H₂, while the applied voltage was varied between 7 kV and 11 kV. Subsequent to igniting the plasma after 5 min, the O₂ mole fraction immediately decreased below 100 ppm. Even full conversion was achieved at applied voltages above 10 kV validating that the complete reactor volume can be treated by the SDBD. However, it is not possible to decouple the chemical activity of the plasma from its influences on the flow characteristics. Therefore, an electrode configuration coated with Pt nanoparticles was taken as a chemically active reference system without plasma. Pt-based catalysts are known to catalyze the reaction of O₂ and H₂ at ambient temperature [40]. Figure 4(b) shows the O₂ mole fraction as a function of time in a gas mixture of 20% H₂ passing over the Pt-coated electrode configuration. At the beginning of the measurement the reactor is bypassed and the gas mixtures directly passes through the online MCA. Subsequent to switching the SDBD reactor online after 10 min, a sharp decrease of the O₂ mole fraction can be observed by flushing N₂ out of the SDBD reactor. After another 10 min a stationary O₂ mole fraction of 470 ppm was established. This qualitative comparison shows that despite the high catalytic activity of Pt-based catalysts only 6% conversion was achieved indicating that the system without plasma is strongly limited by mass transport phenomena.

Zoom In Zoom Out Reset image size

Furthermore, the Reynolds number Re was estimated to obtain a prediction of the flow conditions inside the reaction chamber without plasma using the hydraulic diameter of a rectangular tube. The results are summarized in table 1. It has to be noted that the Reynolds number only provides a rough estimate and that the exact reactor geometry as well as installed parts such as the electrode configuration are not properly considered. As the obtained Reynolds number are far below the critical Reynolds number $R{e_{{\text{crit}}}} = 2300$, it can be assumed that the flow inside the reaction chamber is laminar without plasma and that no significant gas mixing occurs perpendicular to the flow direction.

The low conversion over the Pt-coated electrode configuration as well as the relatively low Reynolds numbers indicate that plasma may change the flow conditions leading to high degrees of conversion. Therefore, the influence of the plasma on the flow characteristics inside the reaction chamber was investigated by Schlieren imaging for a variety of reaction conditions and parameters.

Schlieren images of the SDBD were obtained through the side window of the SDBD reactor in either 100% N₂ or 100% H₂ at a total volumetric flow rate of 10 l min⁻¹. Measurements in the reaction gas mixture of H₂, N₂ and O₂ showed analogous but less pronounced Schlieren structures and are excluded for the sake of simplicity. Figure 5 exemplarily shows Schlieren images of both gas mixtures at 11 kV (approx. 96 W). The SDBD electrode configuration can be seen in the middle of the images. The outlines are defined by the geometry of the side windows corresponding to the inner reactor wall at the top and the bottom and by the viewing angle of the lens at the right and left side. The gas flow was directed from left to right and passed above and below the electrode configuration. The positions of the metal grid lines perpendicular to the flow direction are indicated by vertical lines.

Zoom In Zoom Out Reset image size

For both gas compositions Schlieren structures occurred above and below the electrode configuration covering the complete cross section of the reaction chamber. Even though the shape of the Schlieren structures was roughly comparable for both gas compositions, the intensity and visibility of the Schlieren structures decreased as a function of the H₂ mole fraction, caused by a lower density gradient. This is counterintuitive, since the Gladston–Dale coefficient of H₂ is higher than the one of N₂, which would result in higher Schlieren visibility [41]. However, due to the higher thermal conductivity of H₂, temperature and thus density gradients are less pronounced [42, 43]. As the Schlieren structures were most pronounced for 100% N₂, all subsequent Schlieren images were obtained with this gas composition.

Therefore, a more detailed description of the Schlieren structures is provided for 100% N₂. The starting points of the individual Schlieren structures correspond to the positions of the metal grid lines of the electrode configuration, which are parallel to the line of sight. This observation indicates that the Schlieren structures directly originate from the active plasma volume along the metal grid lines and that their extent is defined by the geometry of the electrode grid. The shape of the Schlieren structures is distinctly different above and below the Al₂O₃ plate. Bright parabolic structures almost resembling quarter circles occur above the electrode configuration and expand from the center of a metal grid line to the right and left side reaching the top of the reaction chamber in between two of the labeled metal grid lines. However, the Schlieren structures on the right side of a metal grid line are more pronounced and superimpose with the Schlieren structure from the following grid line in flow direction. Underneath rather straight Schlieren structures are formed pointing away from the Al₂O₃ plate at an angle of approximately 40°. The most dominant Schlieren structures follow the flow direction and are oriented from left to right reaching the bottom of the reaction chamber below the following metal grid line in flow direction, while Schlieren structures going in the opposite direction are hardly identified.

To obtain a more quantitative description of the Schlieren structures, line profiles for different positions above and underneath the electrode configuration were generated. The line profiles were obtained by plotting the intensity in the visible spectral range of the image along a line parallel to the Al₂O₃ plate as exemplarily shown in figure S3. Three different positions at 20%, 50%, and 80% of the height above or underneath the electrode configuration were chosen. Figure 6 shows the corresponding line profiles, which are in good agreement with previous observations. Above the Al₂O₃ plate the peak positions shift only slightly along the flow direction for the middle and most distant position in comparison to the nearest position, which corresponds to the parabolic shape of the Schlieren structures. In contrast, the intensity maximum below the electrode configuration is shifted in the flow direction, which is in good agreement with the linear Schlieren structures. Furthermore, it has to be noted that the Schlieren intensity of the peak and consequently the density gradient do not reveal a clear trend.

Zoom In Zoom Out Reset image size

For a more detailed evaluation Schlieren images were recorded for applied voltages between 7 kV (approx. 17.5 W) and 11 kV (approx. 96 W) at a pulse repetition frequency of 4000 Hz. A selection of these images is shown in figure 7. It has to be noted that all recording parameters such as the exposure time of the camera or the amount of light cut-off at the knife edge of the Schlieren aperture were kept constant during the variation of the applied voltage. Therefore, the obtained images and line profiles may be compared with regard to relative changes. It is obvious that the extent to which the Schlieren are visible is a function of the applied voltage and consequently of the dissipated power. At 7 kV only weak density gradients can be observed close to the surface of the Al₂O₃ plate. For applied voltages above 8 kV the shape of the Schlieren structures is identical to the one described in the previous section, but the visibility and intensity increases from 7 kV to 10 kV, while it is rather constant above 10 kV.

Zoom In Zoom Out Reset image size

These observations are supported by the corresponding line profiles at the most distant position above and below the electrode configuration (figure 8). Because the underlying trends are more prominent at the most distant position, in the following only these data are shown. The position of the peaks barely changes emphasizing the identical shape of the Schlieren regardless of the applied voltage. Only at 11 kV the peak maximum appears to be slightly shifted against the flow direction. On the other hand, the intensity of the peaks does not reveal a clear trend and the differences in intensity are rather negligible.

Zoom In Zoom Out Reset image size

Another important parameter is the volumetric flow rate, which can influence the flow characteristics inside the reaction chamber as well as the performance of the SDBD in the oxidation of VOCs or other gas purification applications. Therefore, Schlieren images were recorded for volumetric flow rates between 5 l min⁻¹ and 20 l min⁻¹ as shown in figure 9.

Zoom In Zoom Out Reset image size

The shape and extent of the Schlieren structures on both sides of the electrode configuration was changed by the variation of the volumetric flow rate. Nevertheless, the Schlieren structures still covered the complete cross section of the reaction chamber. At a volumetric flow rate of 5 l min⁻¹, the structures above and below the electrode configuration were more symmetric, as rather parabolic Schlieren structures can be observed for the bottom side as well, resembling the ones above the electrode configuration. An increase of the volumetric flow rate to 10 l min⁻¹ or higher leads to the Schlieren shapes previously described. The Schlieren structures are distorted in the flow direction as a function of the volumetric flow rate, and rather straight Schlieren structures are obtained below the electrode configuration, which can be explained by the higher mean velocity of the gas. However, it has to be noted that the distortion along the flow direction also occurs above the electrode configuration, even though it is more pronounced on the bottom side of the electrode configuration. The parabolic shapes on the top side were maintained for all flow rates, but the point where they reached the inner reactor wall was shifted slightly in the flow direction.

The corresponding line profiles at the most distant position above and below the electrode configuration are in good agreement with the previous description of the Schlieren structures (figure 10). The peaks are slightly shifted in the flow direction to the left for volumetric flow rates of 15 l min⁻¹ and 20 l min⁻¹. The most notable deviation can again be observed below the electrode configuration for a volumetric flow rate of 5 l min⁻¹, where the peak maximum is closer to the originating metal grid line. This correlates with the rather parabolic Schlieren structure.

Zoom In Zoom Out Reset image size

To allow better comparability with experimental Schlieren imaging, N₂ was used for all theoretical flow simulations. Figure 11 shows the velocity profile at the center of the reactor for the 2D fluid flow simulation without the plasma discharge model. It can be observed that the flow follows a laminar profile for flow with and without heat transfer effects. The increase in mean velocity in flow with heat transfer model is due to the decrease in fluid density. It can be concluded from the velocity vector of fluid flow with heat transfer as shown in figure 12, that there is no vortex or eddy formation due to the presence of the electrodes and heat transfer effects.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 13 shows the velocity vector of fluid flow for the plasma discharge model. It is observed that due to the electrohydrodynamic force, an accelerated flow is generated on both sides of the electrode, which imparts momentum to the fluid resulting in vortex formation. On the left side of the electrode, the body force generated by the plasma discharge accelerates the fluid away from the electrode against the bulk fluid flow direction and towards the free stream. Thus, a strong curvature in the fluid flow observed which is represented by the velocity vector in figure 13. It is due to the opposite direction of inertial force and the electrohydrodynamic force, which eventually evolves into a vortex structure. On the right side of the electrode, the generated body force due to plasma discharge accelerates the fluid away from the electrode towards the free stream, in the same direction as the bulk fluid flow. Since the direction of inertial force and electrohydrodynamic force are in the same direction, no vortex-like rotational structure is found.

Zoom In Zoom Out Reset image size

The synthetic Schlieren image was generated using PARAVIEW by calculating the gradient of the density and comparing it with the experimental Schlieren images. The Schlieren structures generated along the x- and y-direction are shown in figures 14 and 15, respectively. It can be observed that the density gradient along the x-direction closely resembles the Schlieren image obtained from experiments with respect to the position of dark and light shade regions. For the top side of the electrode configuration in figure 14, density gradients of high intensity are observed above the configuration indicating a sharp change in density from top to the sides of the electrode. This high density gradient can be attributed to the movement of unreacted species from the bulk fluid to the reactive plasma region of low density. Density gradients of low intensity are observed on both sides of the configuration, which can be attributed to the movement of highly reactive species due to the electrohydrodynamic force generated by the plasma discharge away from the electrodes. Thus, the density gradients observed is a combined effect of density change due to plasma discharge process accompanied by the strong electrohydrodynamic forces along with the gas heating effects. Comparing the synthetic Schlieren images with the experimental ones shows that the position of the dark and light shade regions are similar, but in the experiment to a lesser extent. The experimental Schlieren images show that the light shade region extends until half the distance to the adjacent gridline, whereas the light shade region in the simulated Schlieren images covers less distance and the effect is observed most strongly close to the electrode. This deviation in the extent of the plasma discharge could be due to the cumulative effect of the electrohydrodynamic force occurring in a 3D structure in all directions, which is not completely described by modeling the plasma through a simplified phenomenological model.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

While the deposition of residues is often an adverse side effect, it can also yield valuable information about underlying phenomena. Deliberate deposit formation can be used to gain information about flow conditions and transport phenomena similar to PIV. Thus, we used the intentional toluene decomposition to qualitatively visualize the radial flow behavior induced by the plasma discharge. A gas flow (60% H₂, 40% N₂, 1000 ppm O₂, 10 slm) additionally containing 200 ppm of toluene was fed into the reaction chamber where a plasma was ignited (20 W–90 W) for several hours on stream. An alumina plate was positioned at the reactor lid onto which a viscous, tar-like yellowish substance was deposited. A photo of the alumina plate before and after the plasma ignition is shown in figure 16.

Zoom In Zoom Out Reset image size

Figure 16 shows an emerging grid pattern, which corresponds well to the metal grid of the electrode configuration. The ordered deposition of carbonaceous residues originating from toluene decomposition in the plasma clearly confirms radial mass transport from the electrode configuration to the reactor lid. The movement of the particles is induced by the force of the bulk gas as shown by the simulations. Charging individual particles or droplets of the deposit and subsequent acceleration by the electric field seems unlikely, because it would create this grid pattern only above the powered side. However, this pattern also emerged on the bottom of the reactor without a potential difference relative to the reactor walls (not shown).