**Electrical Discharge in Water Treatment Technology for Micropollutant Decomposition**

Patrick Vanraes, Anton Y. Nikiforov and Christophe Leys

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/61830

#### **Abstract**

Hazardous micropollutants are increasingly detected worldwide in wastewater treatment plant effluent. As this indicates, their removal is insufficient by means of conventional modern water treatment techniques. In the search for a cost-effective solution, advanced oxidation processes have recently gained more attention since they are the most effective available techniques to decompose biorecalcitrant organics. As a main drawback, howev‐ er, their energy costs are high up to now, preventing their implementation on large scale. For the specific case of water treatment by means of electrical discharge, further optimi‐ zation is a complex task due to the wide variety in reactor design and materials, dis‐ charge types, and operational parameters. In this chapter, an extended overview is given on plasma reactor types, based on their design and materials. Influence of design and ma‐ terials on energy efficiency is investigated, as well as the influence of operational parame‐ ters. The collected data can be used for the optimization of existing reactor types and for development of novel reactors.

**Keywords:** electrode configuration, electrohydraulic discharge, energy yield, organic degradation efficiency, dielectric barrier discharge

### **1. Introduction**

In this introductory section, the current status and limitations of both conventional and advanced water treatment systems are explained in detail, with a focus on their performance on micropollutant removal. This context is necessary to understand the role that plasma technology can play in the future challenges of water purification, which is the main topic of this chapter. Further, it opens a clear perspective on research that still needs to be done in order to turn the plasma treatment of water into a mature technology.

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

With ongoing improvement of chemical analytical methods, a wide spectrum of compounds and their transformation products are increasingly detected in water bodies and sewage sludge. Many of these compounds occur in low concentrations in the range of microgram to nanogram per liter. They are therefore called micropollutants. Among these are food additives, industrial chemicals, pesticides, pharmaceuticals, and personal care products. Even with such low concentrations, various environmental effects have been observed. Continuous release of antibiotics in the environment leads, for example, to increasing resistance of microorganisms [1]. Chronic exposure of aquatic life to endocrine disruptive compounds causes feminization, masculinization, and immunomodulating effects in fish and frogs, which has tremendous effects on the ecosystem [2, 3]. Additionally, there is growing concern for the direct acute and chronic effects of micropollutants on human health and safety [4–8]. Although more insight is gained on the impact of individual micropollutants, their synergistic, additive, and antago‐ nistic effects are still vastly unknown. Some countries or regions have adopted regulations for a small number of compounds. Nevertheless, effluent limitation guidelines and standards do not exist for most micropollutants.

The primary source of many micropollutants in aquatic systems is the effluent of conventional wastewater treatment plants. An example of such treatment plant is schematically shown in Figure 1. The treatment process occurs in sequential steps:


removal can be achieved with sand filtration, biological nutrient removal, adsorption on activated carbon, membrane processes, and advanced oxidation processes [9, 12]. However, many conventional wastewater treatment plants have no or limited tertiary treatment, due to the high costs associated with most of these methods.

As should be noted, organic pollution that enters a wastewater treatment plant has three outlets: effluent water, excess sludge, and exhaust gas in the form of CO2 or volatile organic compounds. Excess sludge disposal is complex since it contains a high concentration of harmful substances and only a small part of it consists of solid matter. Sludge treatment is required to reduce the water and hazardous organic content and to render the processed solids suitable for reuse or final disposal. Therefore, sludge is processed in sequential steps of thickening, digestion, and dewatering before disposal [10].

**Figure 1.** Simplified scheme of a conventional wastewater treatment plant, adapted from Flores Alsina and Benedetti et al. [10, 13].

Up to now, conventional treatment plants cannot sufficiently remove micropollutants, as shown by many studies worldwide [14]. Several options are available to improve the elimi‐ nation of these contaminants, including:


Preventive measures will always be limited by the increased demand for industrial, pharmaceutical, and personal care products. Moreover, highly stable micropollutants such as the pesticide atrazine are detected even several years after the discontinuation of their use. Hence, more effective water treatment processes are required. Existing processes in water treatment plants can be optimized by increased sludge ages and hydraulic reten‐ tion times in conjunction with nutrient removal stages and the varying redox conditions associated with them. Temperature and pH control can also enhance micropollutant removal. According to the reviews by Jones et al. and Liu et al., this is expected to be the most economically feasible approach to increase overall water treatment plant perform‐ ance [15, 16]. While such measures indeed lead to more effective removal of micropollu‐ tion in general, they often have limited or negligible effect on several specific persistent micropollutants, as shown in the extended review by Luo et al. [14].

The removal difference among different compounds can be partly ascribed to micropollu‐ tant properties. For example, Henry's law constant as a measure of compound volatility gives an indication how easily a contaminant will be removed by aeration and convec‐ tion. In activated sludge processes, the solid–water distribution coefficient *K*d has been proposed as a relative accurate indicator of sorption behavior [17, 18]. It is defined as the partition of a compound between the sludge and the water phase and takes into account both hydrophobicity and acidity of the molecule. The biodegradation of micropollutants is more complex to predict, but compound structure is an important indicator. Highly branched side chains and sulfate, halogen, or electron withdrawing functional groups generally make a contaminant less biodegradable. Also, saturated or polycyclic com‐ pounds show high resistance to biodegradation [19, 20].

In order to remove such persistent compounds, an additional secondary or tertiary treatment step for wastewater is required. As a cost-effective alternative, the treatment of hospital and industrial effluent can solve the problem at its source. For these purposes, several advanced treatment techniques have been proposed in recent years. Each removal option has its own limitations and benefits in removing trace contaminants. A complete review on this topic was conducted by Luo et al. [14]. Advanced biological treatment techniques are activated sludge, membrane bioreactors, and attached growth technology. As in the case of conventional biological treatment, these methods are commonly unable to remove polar persistent micro‐ pollutants. Coagulation–flocculation processes yield ineffective elimination of most micro‐ pollutants. In contrast, electrochemical separation treatment methods such as electrocoagulation and internal microelectrolysis give very good results, as discussed by Sirés and Brillas [21]. Activated carbon, nanofiltration, and reverse osmosis are generally also very effective for removal of trace contaminants. However, they are associated with high energy costs and high financial costs. Moreover, all these separation methods have the additional problem of toxic concentrated residue disposal, as they only remove the compounds without further decomposition into less toxic by-products. Advanced oxidation techniques, on the other hand, are able to oxidize toxic compounds to smaller molecules, ideally with full mineralization to CO2 and H2O [22]. Nevertheless, their weakest point is as well their high energy demand. Additionally, they can be hazardous to the environment, by their unwanted production of CO2, long-living oxidants, and potentially toxic oxidation products. Therefore, research needs to focus on optimizing existing systems in order to overcome these problems. Optimal water treatment schemes will eventually be decided upon, achieving effluent limitations set by national or international environmental regulations at a reasonable cost. In their extended reviews, Jones et al. and Liu et al. conclude that it seems unpractical for activated carbon, nanofiltration, reverse osmosis, and advanced oxidation techniques to be widely used in conventional wastewater treatment [15, 16]. However, with more and more shortages of drinking water all over the world, the recycling of wastewater treatment plant effluents as a drinking water source seems just a question of time. For such purpose, these technologies may be advantageous due to their high removal efficiency. Moreover, with the recent developments in energy harvesting in wastewater treatment [23, 24], one needs to consider the possibility to make wastewater treatment plants self-sustainable in power consumption, even with addi‐ tional implementation of advanced treatment. As mentioned above, the treatment of hospital and industrial effluent can be a cost-effective approach as well. This last approach has gained a lot of attention over the past several years. For an overview of research done in this field for the case of hospitals, the reader is referred to the reviews of Verlicchi et al. [25, 26].

Comninellis et al. propose in their perspective article a strategy for wastewater treatment, depending on the water's total organic content, biodegradability, toxicity, and other physico‐ chemical requirements, such as transparency [27]. According to this strategy, the use of costeffective biological treatment is only advised when total organic content is high enough. If such wastewater is not biodegradable, advanced oxidation can be used as pretreatment step, to enhance biodegradability, and reduce toxicity. One needs to consider, however, that presence of oxidant scavengers can sabotage the oxidation efficiency. Therefore, the used oxidation technology needs to be geared toward the wastewater under treatment. In the case that total organic content is low, wastewater possesses little metabolic value for the microor‐ ganisms. Then, advanced oxidation technology that effectively mineralizes the targeted pollution can be applied as a one-step complete treatment method. Alternatively, separation treatment can be applied prior to advanced oxidation, where pollutants are transferred from the liquid to another phase and subsequently posttreated.

Following this line of thought, advanced oxidation techniques take a promising place in the quest for micropollutant removal, as they appear the most effective methods for the decom‐ position of biorecalcitrant organics. However, in this time of a growing energy crisis and concerns over global warming, removal efficiency should not be the only objective. Sustainable development on the whole must to be considered. Therefore, the main objective of research on advanced oxidation technology should be optimization in terms of energy cost and effluent toxicity as well as its compatibility with biological treatment.

Examples of advanced oxidation processes are ozonation, hydrogen peroxide addition, chlorination, Fenton process, UV irradiation, radiolysis, microwave treatment, subcritical wet air oxidation, electrochemical oxidation, homogeneous and heterogeneous catalytic oxidation, ultrasonication, and combinations thereof, such as peroxonation, photocatalysis, and electro-Fenton process.

One type of oxidation method is typically insufficient for micropollutant removal, while a combination of oxidation methods with each other or with other advanced treatment techni‐ ques leads to significant improvement up to complete removal, as concluded in many reviews, such as for antibiotics [28], for pharmaceuticals [29], for UV-based processes [30], and for the general case [31, 32]. Subcritical wet air oxidation is not feasible for micropollutant removal [33].

Oturan and Aaron conclude from their review that chemical methods such as Fenton's process and peroxonation perform worse than photochemical, sonochemical, and electrochemical advanced oxidation techniques [32]. For the latter three technologies, they further conclude the following:


In conclusion, a synergetic combination of multiple oxidants and oxidation mechanisms is recommended for efficient micropollutant decomposition. Therefore, water treatment by means of plasma discharge takes an interesting and promising place among the advanced oxidation techniques, as it can generate a wide spectrum of oxidative species and processes in proximity of the solution under treatment, including shock waves, pyrolysis, and UV radia‐ tion. The hydroxyl radical OH is often named as the most important oxidant, due to its high standard oxidation potential of 2.85 V and its unselective nature in organic decomposition. Further, plasma in contact with liquid can generate significant amounts of O3 and H2O2. These two oxidants are frequently used in other advanced oxidation methods and lead together to the peroxone process. Other important reactive oxygen plasma species include the oxygen radical O, the hydroperoxyl radical HO2, and the superoxide anion O2 – . When electrical discharge occurs in air, also reactive nitrogen species such as the nitrogen radical, the nitric oxide radical NO, and the peroxynitrite anion ONOO– will play an important role. Oxidants can either enter the liquid phase through transfer from the gas phase or be formed directly in the liquid phase at the plasma–water interface by interaction of plasma species with water or dissolved molecules. Radical hydrogen H, a powerful reducing agent, is directly formed in aqueous phase by the electron collision with water molecules. Plasma discharge also leads to aqueous electrons, which even have a stronger reduction potential. A more detailed overview of oxidative plasma species and processes without nitrogen-containing oxidants was conduct‐ ed by Joshi and Thagard [34]. Recent insights on the chemistry of plasma-generated aqueous peroxynitrite and its importance in water treatment are given by Brisset and Hnatiuc and Lukes et al. [35, 36].

The presence of different oxidants reduces the selectivity of an oxidation method. Direct oxidation of organics by ozone is, for example, very selective. This is illustrated with the reaction rate constants listed by Jin et al., Sudhakaran and Amy, and Von Gunten [37–39], which range from 10–5 to 3.8 × 107 M–1s–1. Hydroxyl, on the other hand, is considered unselective in the decomposition of organics, with reaction rate constants from 2.2 × 107 to 1.8 × 1010 M–1 s–1, as listed in the same 3 references and [40]. Hence, organic oxidation by means of plasma is expected to be rather unselective as well. Nevertheless, the chemistry behind plasma treatment is very complex due to the interactions between the various reactive species in the gaseous phase, in the aqueous phase and at their interface. Moreover, this chemistry is strongly dependent on the used electrode configuration and material, discharge regime, applied voltage waveform, water properties, and feed gas. Therefore, the optimization of plasma reactors for water treatment in terms of energy costs and effluent toxicity is a complicated task, which still requires more research effort and insight. On the positive side, plasma-based water treatment has already shown itself as a versatile technology, which can find application in the treatment of biological, organic, and inorganic contamination, after sufficient optimization has been reached. As an additional advantage, its flexible design allows it to be easily combined with other advanced treatment techniques. Such combinations can lead to interesting synergetic effects and further optimization.

As pointed out already by Sillanpää et al. [41], no information has been reported in scientific literature on the treatment cost of advanced oxidation processes. To our knowledge, plasma technology has only been compared with other water treatment methods in the review of Sirés and Brillas [21]. According to their review, the main drawback of plasma technology would be its high energy requirement. However, the authors did not validate this claim with quantitative data. Comprehensive quantitative assessment is needed to compare different techniques better from both economic and technical points of view, as Luo et al. conclude in their review [14]. For the specific case of plasma technology for water treatment, only 3 reviews have extensively compared different reactors in their energy efficiency for organic decompo‐ sition. Malik was the first to do this for dye degradation [42]. Recently, Bruggeman and Locke and Jiang et al. have made a comparison for mostly phenolic compounds [43, 44]. Their findings will be discussed in next sections. Nevertheless, the main comparative parameter in these reviews is energy yield G50 for 50% pollutant removal, expressed in g/kWh, which is only used in literature on plasma treatment. Moreover, G50 is strongly dependent on the initial pollutant concentration. It is therefore not fit for comparison with other advanced treatment techniques.

In this chapter, several plasma technologies are discussed that can be applied in water treatment. First, different approaches for reactor classification will be analyzed in detail. In Section 3, an overview will be given of different plasma reactor types, where classification is based on reactor design and reactor materials. The influence of multiple working parameters on reactor energy efficiency is discussed in Section 4. Section 5 gives a summary with future prospects and concluding remarks on the research that is still required in this field.

### **2. Approaches for reactor classification**

Plasma reactors for water treatment can be classified in many ways, depending on several criteria. Such classifications have already been made in a few reviews. One popular approach starts from 2 or 3 main plasma–water phase distributions and subdivides reactors further based on their electrode configuration, as in Bruggeman and Leys and Locke et al. [45, 46]. Here, we will adapt this approach with 6 plasma–water phase distributions, where electrical discharge is generated


Accordingly, these reactor types are called (1) electrohydraulic discharge, (2) bubble discharge, (3) gas phase discharge, (4) spray discharge, (5) hybrid, and (6) remote discharge reactors. The basic idea behind this classification is that the total plasma–water interface surface is an important, determining parameter for a reactor's energy efficiency. A larger interface surface is expected to cause higher pollutant degradation efficiency, in agreement with the reviews of Malik, Jiang et al., and Bruggeman and Locke [42–44]. Interface surface can be enlarged by generating plasma in bubbles, by spraying the solution through the active plasma zone, and by making the solution flow as a thin film along the discharge. It can be further enlarged by extending the plasma volume, which has led to many possible choices of electrode configu‐ rations and geometries. Additionally, dielectric barriers and porous layers are often introduced in the setup to avoid unwanted energy losses to Joule heating of water and spark formation, while enhancing the local electric field for easier breakdown. In Section 3, this approach for reactor classification will be studied in more detail.

Another popular approach classifies reactors based on the used discharge regime and applied voltage waveform. The following discharge regimes should be considered:


Microdischarge and Townsend discharge are not included in this list, as they are not used for water treatment plasma reactors to our knowledge, except in ozone generation. Spark discharge is to be understood as a transient form of arc discharge and therefore falls in the fourth category. All four discharge regimes can be formed in the gas phase or the liquid phase, although we only found one very short and recent report on submerged DBD discharge without bubbles for water disinfection [47]. In the gas phase, corona discharge needs lowest power input and arc discharge the highest. Underwater discharge requires additional energy for plasma onset by cavitation, but it has the advantage of a large plasma–liquid contact surface. Locke et al. and Jiang et al. suggested in their reviews to apply low-energy plasma such as corona and glow discharge for the treatment of water with low contaminant concen‐ tration. On the other hand, high-energetic arc discharge might be more effective for high pollutant concentration [44, 46].

Depending on the voltage waveform used, the following types of discharge can be distin‐ guished:



Variations are possible, such as periodically interrupted AC discharge to avoid excessive heating of the plasma gas [48]. Multiple voltage-related parameters can influence reactor energy efficiency, such as voltage amplitude and polarity, sinusoidal or pulse frequency, and pulse rise time and width. Their influence will be discussed in more detail in Section 4. By combining different voltage waveforms with the different discharge regimes in either the liquid or the gas phase, a list of discharge types can be obtained. Malik, for example, identified 27 distinct reactor types in his literature study, using this approach [42]. Locke et al. made a classification into 7 types [46]:


This classification partly overlaps with the 6 types of plasma–water phase distribution mentioned above, where bubble discharge is included in hybrid gas–liquid discharge. Jiang et al. add 4 more types [44]:


While the first 3 types generate plasma in or in contact with the liquid, the last type introduces the water under treatment directly in the torch, where it subsequently gets vaporized and becomes the plasma forming gas. It can therefore be classified as a subtype of spray discharge reactors. For more summarized information on the basic chemistry and physics of the dis‐ charge types in this list, the reader is referred to [43, 46].

The list illustrates how the 6 reactor types based on plasma–water phase distribution can be further split into subtypes. This has been done in a more extensive way by Locke and Shih, who identified more than 30 subtypes during their comparative study for the reactor energy efficiency of H2O2 production [49]. While this method has proven to be useful for their purpose, it disregards the influence of reactor design, materials, and working parameters. As Locke and Shih point out, studying this influence is a challenging task, but it will reveal valuable information on reactor optimization.

A small number of reviews have been made on plasma reactor energy efficiency for organic decomposition, which mostly focused on the influence of plasma–water phase distribution, discharge type, and a few working parameters. Nevertheless, no comprehensive overview is found in literature up to now on reactor design and reactor materials. The next section will deal with this topic. While the section is not meant to be fully comprehensive, we want to give a broader overview of reactor types reported in literature than has been done in prior reviews. Such overview is not only useful to get easier and faster understanding of reactor operation and development but also serves as source of inspiration for future reactor designs. Moreover, it reveals the close relationship between electrode configuration and discharge type. Both popular and more exotic, unique designs will be discussed, with extra focus on several general working parameters in Section 4. This review also includes plasma reactors that have been used for inorganic removal or biological treatment, as these reactors are also valuable candi‐ dates for aqueous organic decomposition.

### **3. Overview of reactor types**

#### **3.1. Electrohydraulic discharge reactors**

Electrohydraulic discharge reactors have been studied for many years due to their importance in electrical transmission processes and their potential for water treatment. From theoretical point of view, they are attractive for water treatment due to the relatively high ratio of plasma– water contact surface to plasma volume and proximity of plasma to the water surface. Moreover, they generate shock waves that can aid in organic decomposition. Nevertheless, electrohydraulic discharge reactors are usually less efficient than other reactor types for water decontamination [42, 50]. This is likely due to the additional input energy required for cavitation, i.e., gas phase formation during discharge onset.

Most commonly reported types of electrohydraulic discharge are pulsed arc and pulsed corona discharge. To our knowledge, all arc electrohydraulic discharge reactors reported in literature have pulsed input power. Mostly, a rod-to-rod electrode configuration is used (Figure 2a), but there are also reports on a reactor with a grounded L-shaped stationary electrode and a vibrating rod electrode (Figure 2b) [51, 52]. As learned from personal communication with Dr. Naum Parkansky, the vibrating electrode has the purpose to facilitate electrical breakdown and mix the treated solution. Apart from organic degradation, pulsed arc electrohydraulic discharge is also gaining more recent attention for biological treatment [53, 54]. Often, refractory metal such as tantalum, titanium, tungsten, or a corresponding alloy is selected as electrode material, as it needs to be sufficiently resistant to corrosion and shock waves. Tungsten has proven to be less corrosive than titanium and titanium alloy [53]. Particles that eroded from titanium electrodes enhanced methylene blue decomposition during aging in one study. The authors explained this enhancement with titanium peroxide formation from interaction of H2O2 with the particle surface [55]. In contrast, particles that eroded from low carbon steel electrodes diminished the decomposition of the same pollutant, possibly through the catalytic decomposition of H2O2 and scavenging of OH radicals [52]. While an increasing energy efficiency is reported for the decomposition of 2,4,6-trinitrotoluene [56] and methyltert-butyl ether [57] when the electrode gap is reduced, the opposite effect has been observed for atrazine degradation in another study [58]. Further investigation is needed to understand this apparent contradiction. In a small comparative study, Hoang et al. found the energy efficiency of 4-chlorophenol decomposition of this type of reactor to be one order of magnitude lower than degradation efficiency with UV, UV/H2O2, and O3 systems [59].

Submerged pin-to-pin electrode configuration, as in Figure 2c, is uncommon in scientific literature. Such reactor has been used with high frequency bipolar pulsed power with reduced voltage and low pulse energy by Potocký et al. [61], to lower temperature loading of the electrodes and to clean both electrode tips continuously from any possible adjacent products. With addition of a high inductance in series with the discharge, transition from glow type to arc type discharge can be suppressed [62]. For sufficiently large interelectrode distance and low voltage amplitude, unbridged nonarc discharge was observed. For closer electrodes and higher voltage, bridged arc discharge was obtained, which, according to the authors, was initiated with spark formation at both electrodes [63].

A pin-to-plate electrode configuration (Figure 2d) is often used in electrohydraulic discharge reactors, either for DC glow [64, 65] or pulsed corona [66–68] discharge with positive polarity. The pin curvature radius determines the local electric field strength and is therefore an important parameter that influences discharge initiation [69]. For pulsed corona discharge, the anode pin material has been reported to lead to catalytic effects for organic decomposition.

**Figure 2.** Types of electrohydraulic discharge reactors: (a) pulsed arc, (b) pulsed arc with vibrating electrode, (c) pin-topin, (d) pin-to-plate, (e) multi-pin-to-plate, (f) brush-to-plate, (g) plate-to-plate with porous ceramic coating, (h) coaxial rod-to-cylinder with ceramic coating on rod, (i) coaxial wire-to-cylinder with ceramic coating on wire, (j) diaphragm discharge, (k) capillary discharge, (l) coaxial diaphragm discharge reactor from Šunka et al. [60] with perforations in tubular electrode covered by polyethylene layer, (m) contact glow discharge electrolysis, (n) RF-discharge in cavitation bubble on electrode, (o) microwave bubble plasma from waveguide with antenna slot, and (p) "hot spot" plasma for‐ mation on activated carbon surface under influence of microwave irradiation.

Platinum enhances pollutant degradation as compared to a NiCr, but only in combination with certain electrolytes. In the case of ferrous salts, this is due to reduction of Fe3+ to Fe2+ by erosion particles from the platinum electrode [66, 67]. Erosion of tungsten, on the other hand, causes catalysis of oxidation by plasma-generated H2O2 [70].

To achieve a higher plasma volume, the pin electrode can be replaced by a multipin electrode [71], a brush electrode [72], or a plate electrode coated with a thin ceramic layer [73], respec‐

tively, shown in Figures 2e–g. The concentration of predischarge in the pores of the ceramic layer enhances the electric field strength on the electrode surface [68]. Due to inhomogeneities such as entrapped microbubbles inside the ceramic layer, the electric field can be locally even higher. As a result, a large number of streamers can be generated with lower input voltage as compared to uncoated electrode systems. Moreover, a ceramic coating does not only facilitate an upscale of the system but also serves as a support for a suitable catalyst of the plasma chemical reactions. Also, certain ceramic materials can enhance organic decomposition by catalytic effects [74]. Usually, positive high-voltage pulses are applied to the coated electrode. In some cases, negative high voltage is used to avoid arc formation, as this can damage the coating [73]. However, bipolar pulses are advised since monopolar pulses cause a polarized charge buildup on the ceramic, which can quench the electrical discharge [75].

To enlarge the plasma volume even further, coaxial geometry has been used, where either a coated rod [75] or wire [68, 74] high-voltage electrode is located at the symmetry axis of the grounded cylindrical electrode (Figures 2h–i). Analog to the curvature of pin electrodes, the diameter of the inner electrode is an important system parameter.

When a submerged anode and cathode are separated from each other with a perforated dielectric barrier, electrohydraulic discharge will occur through cavitation at the perforation. For larger ratio of perforation diameter to thickness, this type of discharge is called diaphragm discharge (Figure 2j), while for lower values, the term capillary discharge is used (Figure 2k). DC glow discharge [76, 77] and pulsed corona discharge [78] are commonly used for this reactor, but also AC power input is possible. The strongly inhomogeneous electric field during plasma onset has a similar structure to the one in pin-to-plate geometry. Therefore, similar plasma features can be expected. Energy efficiency for dye and phenol degradation with a single diaphragm is the same as in the case of a pin-to-plate electrode system [46]. Also, similarities with contact glow discharge electrolysis are reported [77]. As an important difference, diaphragm discharge is not in direct contact with the electrodes, which prevents electrode erosion [76]. Sunka et al. developed a coaxial reactor where a polyethylene covered tubular anode with perforations was placed inside a cylindrical cathode (Figure 2l) [60]. The generated plasma was reported to be similar as well.

Another common type of electrohydraulic discharge is contact glow discharge electrolysis. As depicted in Figure 2m, a pointed anode is placed with its tip in the water surface. It is separated from the submerged cathode by means of a sintered glass barrier. In such reactor, glow discharge is generated at the anode tip in a vapor layer surrounded by water. Plasma volume can be increased by increasing the anode number. Stainless steel performed better as anode material than platinum for Acid Orange 7 decoloration [79].

Electrohydraulic discharge can also be generated with RF or microwave power, but such reactor types are less common. Figure 2n shows a reactor where plasma is generated in a cavitation bubble on the tip of an RF electrode [80]. Producing cavitation bubbles by means of microwave power is more complicated. Therefore, a slot antenna can be placed in between the liquid and a microwave guide, as illustrated in Figure 2o. The electric field intensity can be enhanced by installing a quartz plate with holes, a so-called bubble control plate, on the slot antenna. Ishijima et al. reported an increase of methylene blue decomposition efficiency with a factor of 20 after installation of the bubble control plate and tripling the amount of slot antennas [81]. Another way to produce underwater plasma with microwave power is by adding a microwave-absorbing material with high surface area, such as activated carbon, to the solution under treatment. Under influence of microwave irradiation, delocalized πelectrons on the activated carbon surface gain enough energy to jump out of the surface and generate confined plasmas (Figure 2p), also called hot spots, which are known to increase organic decomposition efficiency [82].

#### **3.2. Bubble discharge reactors**

Since electrohydraulic discharge generally has low energy efficiency due to the difficulty of initiating discharge directly in the water phase, a lot of attention has gone to enhancing efficiency by discharge formation in externally applied bubbles. Bubbling has the additional advantage of mixing the solution. Moreover, discharge initiation in the gas phase minimizes electrode erosion, which lengthens the lifetime of the system. Bubbling gas through the discharge region greatly increases radical density in the plasma, as for example observed by Sun et al. [69] for O2 and Ar bubbles. Obviously, feed gas plays a determining role. Yasuoka et al. measured highest efficiency for Ar bubbles, in which plasma spread extensively along the inner surface, while lowest efficiency was obtained with He plasma, which has smallest plasma–water contact surface [83]. The importance of the working gas in plasma reactors in general will be further discussed in Section 4.

A common method is to pump gas upward through a nozzle anode, located underneath a grounded electrode [69], as shown in Figure 3a. Often, the nozzle electrode is placed inside a dielectric tube up to its tip, to avoid any energy leakage toward the water. Alternatively, a pin anode is sometimes placed inside a dielectric nozzle which transports the feed gas (Figure 3b) [86]. The pin tip can be placed below or above the nozzle extremity. Many variations can be encountered in literature, such as a pin anode inside a perforation in a dielectric plate (Figure 3c) [87] and different nozzle orientations (Figures 3d–e) [88, 89]. All choices in nozzle or perforation material, shape, dimensions, and orientation determine the bubble shape during formation and its position after detachment, which significantly influences the electric field in the interelectrode region and therefore the plasma characteristics. This complicates compari‐ son of different reactors. Another option is to place the high-voltage electrode underneath the perforated dielectric plate, as shown by Yasuoka et al., Sato et al., and Yamatake et al. [83, 90, 91] (Figures 3f–h), where the ring-shaped grounded electrode is located around the bubble. Also, here, electrode geometry and position influence the electric field. Yasuoka et al. found their single hole reactor (Figure 3h) to be more energy efficient than advanced oxidation with photochemical persulfate, photocatalyst heteropoly acid, photodegradation, and ultrasonic cavitation for the decomposition of 2 surface active compounds [83]. By increasing the number of nozzles or holes, energy efficiency can be enhanced. In a study by Sato et al. [90], a reactor with a single hole (Figure 3f) was compared to a reactor with 9 holes (Figure 3g). Discharge power deposited per hole was lower in the reactor with 9 holes, which seemed to minimize self-quenching of OH radicals, resulting in higher efficiency. Following this line of thought, a multibubble system as in Figure 3i with a high-voltage mesh in the gas phase attached to a

Electrical Discharge in Water Treatment Technology for Micropollutant Decomposition http://dx.doi.org/10.5772/61830 443

**Figure 3.** Types of bubble discharge reactors: (a) upward nozzle electrode-to-plate, (b) upward nozzle containing nee‐ dle electrode-to-plate, (c) hole containing needle electrode-to-plate, (d) downward nozzle electrode-to-plate, (e) down‐ ward nozzle containing needle electrode-to-plate, (f–h) hole above electrode and inside circular ground electrode, (i) multibubble discharge on porous ceramic with high-voltage mesh, (j) multibubble discharge on porous ceramic tube surrounding a high-voltage wire electrode, (k) bubble discharge on hole in grounded electrode, (l–m) bubbles rising toward high-voltage electrode, (n–p) bubbles rising in between electrodes, (q) DBD reactor with inner barrier, (r) DBD reactor with outer barrier, (s) DBD reactor with double barrier, (t) DBD reactor with spiral electrode from Aoki et al. [84], (u) glass bead packed-bed DBD reactor, (v) coaxial arc discharge reactor with rotating multipin electrode from Johnson et al. [85], (w) multielectrode slipping surface discharge reactor, (x) microwave discharge in ultrasonic cavita‐ tion bubble, and (y) stationary bubble under high-voltage mesh.

porous ceramic seems a promising alternative [92]. Similarly, bubbling gas through a porous ceramic tube containing a high-voltage wire electrode [93] (Figure 3j) gives a large plasma– water contact surface as well. Increasing gas flow rate had no influence in this reactor on decomposition efficiency of phenol, while energy efficiency was enhanced for Acid Orange II removal.

Another situation is found when bubbles are formed on the grounded electrode, as shown by Yamatake et al. and Nikiforov [91, 94] (Figure 3k). Reactor from Figure 3h has more aggressive reaction with water and improved durability of electrode in comparison to the reactor from Figure 3k, as concluded by Yamatake et al. [91]. Nevertheless, one needs to take into account the applied voltage type. While for most of the reactors above positive pulsed corona discharge is used, less commonly AC voltage [94] or positive DC voltage [89, 91] is applied, leading to significantly different phenomena. As Yamatake has shown, plasma can be stably generated in the reactor from Figure 3h even without gas flow since O2 gas is generated from electrolysis by application of positive DC voltage, while this is not the case for the reactor from Figure 3k. The microdischarge channel makes plasma more stable than that in the study of Kurahashi et al. [89], where the same phenomenon has been studied. Therefore, the system of Figure 3k requires higher gas flow due to the short lifetime of the oxygen radical, while in the reactor of Figure 3h decomposition did not depend on flow rate due to direct reaction with water.

Gas can also be bubbled from the ground electrode toward a multipin or brush high-voltage electrode, as shown by Chen et al. and Wang et al. [72, 95] (Figures 3l–m), or in between 2 sideways positioned electrodes, as shown by Lee et al., Miichi et al., and Vanraes et al. [96– 98] (Figures 3n–p). The discharge systems are often very similar to the ones described in Section 3.1 without bubbles. For a plate-to-plate configuration, pulsed streamer discharge mostly occurred inside bubbles adjacent to the electrodes [97].

A last common type of bubble discharge reactor has a coaxial DBD geometry. The cylindrical dielectric barrier can either be placed at the inner [99] or at the outer electrode [91], or at both [91] (Figures 3q–s). For a single barrier, a reticulate electrode can be placed in contact with the water in order to enhance the local electric field [99]. With double barrier, unwanted erosion of the electrodes can be avoided. In the study of Yamatake et al. [91], a double-barrier reactor was found to have higher decomposition efficiency of acetic acid in comparison to a single barrier reactor with similar dimensions, which is most likely caused by a difference in energy density. Monopolar pulsed, bipolar pulsed, and AC high-voltage are most commonly applied to the inner electrode. In the study of Aoki et al. [84], however, the reactor from Figure 3t was used to generate RF glowlike plasma in bubbles. The grounded spiral electrode impeded the motion of the bubbles in the small gap between the electrodes, which increases the probability of discharge in the bubbles. Nevertheless, most of the input energy seems to be dissipated as heat in this system, resulting in low energy efficiency. Bubble movement can also be impeded by adding obstacles in the water bulk. For the single barrier reactor of [100] (Figure 3u), the addition of spherical glass beads significantly enhanced the energy efficiency of indigo carmine decomposition. Porous ceramic sphere gave worse performance than glass beads, but better one than inert conductive fragments. As an interesting research question, it is still unclear whether bubbles in contact with the electrodes in these DBD systems give rise to a better efficiency than freely rising bubbles or not.

A more exotic and patented reactor design is investigated by Johnson et al. [85] (Figure 3v). In this reactor, the high-voltage electrode is a pin array that can rotate at speeds up to 2500 rpm. Oxygen is pumped through the stationary electrode and nebulized to form a bubble mist between the electrodes. DC voltage is applied to the system, generating arc discharge. Rotating the electrode distributes erosion particles evenly on the stationary electrode, preventing pitting and unwanted changes in the relative distance between the electrodes. Moreover, it reduces mass transfer limitations that are apparent in pin-to-plate reactors. Additionally, it lowers the inception voltage by abating the effective distance between the pin electrodes and the station‐ ary electrode. Energy efficiency for methyl tert-butyl ether decomposition increased with an increase in spin rate. The system was found to be more energy efficient than many coronabased technologies but still requires further optimization.

Anpilov et al. developed the multielectrode slipping surface spark discharge system depicted in Figure 3w [101]. A series of cylindrical electrodes are mounted on the outside of a dielectric tube. One of the extreme electrodes is grounded, and high-voltage pulses are applied to the other extreme electrode. All other electrodes are on floating potential. To increase system efficiency, the outer surface of the electrodes is coated with a thin insulating layer. Gas is pumped into the electrode gaps through drilled holes in the dielectric tube. When a highvoltage pulse is applied, plasma discharge occurs initially in the first interelectrode gap adjoining the high-voltage electrode. The breakdown of this gap quickly transports the highvoltage potential to the next electrode, which leads to the breakdown of the second gap. The process repeats itself until the grounded electrode has been reached. This way, the discharge load on each electrode can be kept low, enhancing erosion resistance and increasing the system lifetime. The effectiveness of the system has been demonstrated for disinfection of biological wastewater, methane conversion, and aqueous organic waste decomposition [102].

Apart from bubble cavitation by electrical discharge, the classical gas pumping method, and gas formation by electrolysis, bubbles can also be generated with ultrasonic cavitation. Since microwave power cannot easily generate cavitation bubbles, a combination of ultrasound and microwave discharge can be an attractive method. An example is given by Horikoshi et al. [103] (Figure 3x). Stationary bubbles, however, are not commonly used in bubble discharge reactors, unless for diagnostic purposes. An example is given by Yamabe et al. [104], with the reactor of Figure 3y. This reactor was used to investigate plasma formation and propagation along the gas–water interface, which is a common feature for bubble discharge reactors and which makes it, therefore, different in nature than most gas phase discharge reactors.

#### **3.3. Gas phase discharge reactors**

Electrical discharge in the gas phase is usually more energy efficient for organic degradation than discharge in the liquid phase [42, 43, 50]. In this section, we will distinguish 4 subgroups of gas phase discharge reactors: corona and glow discharge over a horizontal water surface (Figure 4), DBD over a horizontal water surface (Figure 5), falling water film reactors (Figure 6), and arc discharge over a water surface (Figure 7).

#### *3.3.1. Corona and glow discharge over water surface*

The most standard version of a discharge over water surface has a pin-to-water configuration with a grounded water electrode, as depicted in Figure 4a. The type of discharge produced in this reactor, corona, glow, or transient glow-to-spark, depends on the applied voltage, pin curvature, interelectrode distance, and voltage polarity [105, 106]. For the application of water treatment, both positive and negative DC and monopolar pulsed voltage have been reported. AC input power is less common but has been used as well [107]. Plasma volume can be increased by replacing the high-voltage pin electrode with a multipin [108], a brush [72], or a horizontal wire [109] (Figures 4b–d). In the study of Miyazaki et al. [108] with a multipin electrode, the energy efficiency for a certain amount of phenol decomposition was independent of the type of discharge, the voltage amplitude, the polarity of the applied voltage, and the amount of pin electrodes.

**Figure 4.** Types of reactors with corona or glow discharge over water surface: (a) pin-to-water, (b) multi-pin-to-water, (c) brush-to-water, (d) wire-to-water, (e) nozzle electrode-to-water, (f) low-pressure pin-to-water glow discharge reac‐ tor, (g) pin above radial water flow, (h) pin above flowing liquid electrode, (i) miniature microjet above flowing liquid electrode, and (j) multipin above water flow.

According to Dors et al., atmospheric pressure glow discharge in air produces gaseous nitrogen oxides, leading to formation of undesirable aqueous nitrates and nitrites, while DC positive corona produces ozone in air without any traces of nitrogen oxides [110]. The energy efficiency of phenol oxidation in their system, depicted in Figure 4e, was comparable to the results obtained in pulsed corona discharge systems. Sharma et al., however, compared their lowpressure negative DC glow discharge reactor (Figure 4f) with bench scale data of atmospheric pressure corona discharge and concluded that the power cost for pentachlorophenol decom‐ position was lower for their system [111]. Additionally, the operating cost of their reactor was found to be comparable with power cost of UV-based advanced oxidation technologies. Yet it is unclear how feasible such low-pressure system is for applications with large water volume or continuous water flow.

It is important to note that dimensions and movement of the water phase in this type of reactors can influence energy efficiency significantly. In the system of Sharma et al., for example, stirring rate increased the rate of pentachlorophenol removal [111]. Water movement also plays an important role in radial flow reactors (Figure 4g). In the reactors from Jamróz et al. [112] with small sized flowing liquid cathode (Figures 4h–i), degradation efficiency depended strongly on the water flow rate, while the Ar flow rate from the miniature flow Ar microjet (Figure 4i) gave negligible effect on methyl red decomposition. The importance of water and gas flow rates will be further discussed in Section 4.3. As mentioned above, making the solution flow as a thin film along the discharge is another way to enhance the oxidation process. Promising results from a pilot-scale system with negative pulsed corona from multiple carbon fiber cathodes above a flowing water film (Figure 4j) have been published by Even-Ezra et al., Gerrity et al., and Mizrahi and Litaor [113–115]. The system was similar or more efficient than a pilot-scale UV/H2O2 advanced oxidation process and achieved similar energy efficiency to those reported in the literature for other advanced oxidation processes [113, 114]. Moreover, the plasma pilot system with additional ozone injection was more cost-effective than three other commercialized advanced oxidation systems (O3/H2O2, O3/UV, and O3/H2O2/UV) [115]. However, the total capital costs and reliability of large-scale gas discharge reactors for water treatment are still relatively unclear.

#### *3.3.2. DBD over water surface*

DBD over horizontal water surface is most commonly powered with AC voltage, but occa‐ sionally pulsed high voltage has been used as well. Often, glass is used as dielectric barrier, especially quartz glass, while Al2O3 ceramic barriers have also been reported less frequently. Interestingly, reactors with DBD over water often have energy efficiency for organic decom‐ position that increases with input power [116–118]. The most standard reactor design for DBD in the gas phase over a water surface is shown in Figure 5a. In the study of Hu et al. [116], energy efficiency in such reactor was found to increase by decreasing the distance between the dielectric barrier electrode and the water surface. This can be explained with a decrease in plasma volume and thus in unused plasma reactions far from the water surface. A water batch can also be placed in between two dielectric barriers to avoid erosion of one of the electrodes, as in the study of Hijosa-Valsero et al. [119] (Figure 5b). A more exotic way of bringing DBD over a water surface is shown in Figure 5c, where water is kept on floating potential. However, this device has not been applied for organic degradation and is instead used for biomedical applications [120]. In another less common design, the water can be located above the dielectric barrier, with an uncovered high-voltage electrode positioned above the water surface, as in the wire-to-water reactor of Marotta et al. [121] (Figure 5d). In this reactor, the energy efficiency of phenol decomposition was 3.2 times higher with stainless steel wires as compared to Ni/Cr wires. However, it is unclear whether this effect should be attributed to the larger diameter of the Ni/Cr wires or their possible inhibiting effect on ozone or other oxidants. The material of the ground electrode in contact with water in many DBD reactors is, on the other hand, clearly important. In the study of Lesage et al. [122], the use of a stainless steel substrate resulted in a better decomposition efficiency in comparison to the use of brass. This is explained with corrosion of the brass substrate under influence of nitrate, leading to formation of aqueous nitrite, which scavenges OH radicals and thus inhibits the degradation process. Stainless steel, as a more inert metal, does not have this effect.

**Figure 5.** Types of reactors with DBD over water surface: (a) DBD-to-water, (b) water batch in between DBD, (c) DBD above floating electrode, (d) wire-to-water DBD, (e) DBD above radial water flow, (f) DBD above water flow, (g) water flow in between DBD, (h) DBD over radial water flow on porous ceramic, (i) DBD over water flow on porous ceramic, and (j) DBD rod to falling water film.

Also for DBD-based systems, the movement of water influences the degradation efficiency. Reactors with radial flow [117] or flowing water films [118, 123] are investigated in literature for organic decomposition (Figures 5e–g). Both situations have also been reported with incorporation of a porous ceramic segment in the zone between electrodes (Figures 5h–i) [124, 125]. This porous segment serves as guide for the flowing water. It allows the water to remain undisturbed by the electrical discharge due to hydrophilic force. This enables a reduction of the discharge gap and subsequently an increase in the intensity, stability, homogeneity, and efficiency of the discharge. Under such a configuration, a transition from filamentary mode to semi-homogeneous mode of the plasma discharge can be realized [125]. Moreover, such ceramic can be effectively used as substrate for photocatalysts, where both substrate and catalyst remain unchanged after use [124].

An exotic reactor type where a rod high-voltage electrode with dielectric cover is placed next to a falling water film has been investigated by Lesage et al. [122, 123] (Figure 5j). The reactor was found to be significantly more efficient than a gliding arc (see Section 3.3.4) over the same falling water film, partly due to less corrosion of the brass substrate in contact with the water.

#### *3.3.3. Coaxial reactors with falling water film*

A relatively common falling water film reactor that does not use DBD is shown in Figures 6a– b and is often referred to with the term wetted-wall reactor. Either a rod [126] or more commonly a wire [127–130] high-voltage electrode is placed along the axis of a grounded cylindrical electrode. The falling water film flows along the inner wall of the cylinder electrode where it comes in contact with streamer or corona discharge. Mostly, positive pulsed power is applied on the inner electrode, but also negative DC [127] has been reported. Usually, corona discharge is formed in such reactors for all voltage waveforms, while spark discharge is undesired due to excessive energy dissipation to Joule heating. To prevent spark formation, it is necessary to have the entire inner wall area covered by the water flow [130]. The choice of gas flow direction is important, as concluded by Faungnawakij et al. [131] for negative DC corona. Experiments showed a downward airflow to be more effective than upward airflow for acetaldehyde degradation. In the study of Sano et al. [127], the energy efficiency of a wettedwall reactor with negative DC voltage applied to the inner wire was calculated to be 3 to 4 times higher than in a wire-to-water corona reactor (Figure 4d) over flowing water with negative DC voltage. For the same reactor, energy efficiency was found to be highest for conditions of a smooth water surface, i.e., for a minimal water flow rate where the flow entirely covers the anode inner wall and for an optimal current, which does not disturb the flow by strong ion wind. There is an optimal wall radius, where decomposition efficiency is maximal. For higher radius, many plasma-generated short-lived radicals cannot reach the water film in time, while for smaller radius, the plasma–water contact surface decreases [127]. Sato et al. compared four kinds of coaxial reactors with falling water film and positive pulsed power. They found phenol most energy efficiently removed with the configuration of Figure 6b [128]. Sealing such reactor seems beneficial for energy efficiency, due to better confinement of the produced ozone [129].

Most falling water film reactors generate plasma by DBD, either with AC or monopolar pulsed high voltage. Pulsed DBD in coaxial configuration using O2 is considered as one of the most efficient electrical discharge systems evaluated because of the large surface area and small electrode distance [50]. Several configurations are possible, but in the most common design, the water film flows over the surface of an inner stainless steel rod electrode placed inside a glass cylindrical vessel which acts as dielectric barrier. The outer electrode can be a metal mesh or a metal painted layer which is located around the vessel. Four versions of this reactor design

**Figure 6.** Types of coaxial reactors with falling water film: (a) wetted-wall reactor with rod electrode, (b) wetted-wall reactor with wire electrode, (c–f) 4 variations of falling water film DBD reactor with outer barrier, (g) falling water film DBD reactor with inner spiral electrode, (h) falling water film on glass fiber fabric in DBD reactor with outer barrier, (i) wetted-wall DBD reactor with wire electrode, (j) wetted-wall DBD reactor with double barrier, and (k) configuration of coaxial whirlpool reactor of [132].

are found in literature, where the inner electrode is either grounded or connected to the high voltage and with an upward or downward gas flow (Figures 6c–f). With the reactor of Figure 6f, energy efficiencies of micropollutant decomposition are about one order of magnitude higher than for a water batch in between a DBD reactor (Figure 5b) [119]. A slightly different configuration was used by Ognier et al. [133], where a tungsten wire that was rolled around a dielectric rod served as inner grounded electrode (Figure 6g). In this study, volatile aqueous compounds were treated. The more volatile the compound was, as expressed with the Henry's law constant, the more efficiently it was removed. Therefore, degradation processes of pollutants in the gas phase should be considered in plasma reactors, depending on the volatility of the compound. In the study of Bubnov et al. [134], the inner electrode of a coaxial DBD reactor was covered with a 1-mm-thick porous hydrophilic glass-fiber fabric (Figure 6h). This fabric allows a more homogeneous water flow and higher water retention time. Moreover, it can function as substrate for catalysts, such as Cu and Ni compounds, which enhanced the decomposition efficiency in the research of Bubnov et al. [134].

Less frequently, the water film is chosen to flow along the inner wall of the dielectric barrier. Morimoto et al., for example, investigated the effect of placing a dielectric barrier inside the wetted-wall reactor of Figure 6b, as shown in Figure 6i. Addition of the barrier allows to decrease the interelectrode gap without formation of spark discharge, which is expected to increase energy efficiency. With the application of positive nanosecond pulsed high voltage on the inner wire, the treatment efficiency of the DBD system was found to be, surprisingly, less energy efficient for indigo carmine decomposition as compared to the normal wetted-wall reactor. Another type of wetted-wall DBD reactor with falling water film is investigated by Rong et al. [135], for a reactor with double dielectric barrier (Figure 6j). A more exotic type of DBD reactor with modified water-gas mixing is reported by Chen et al. [132]. The system is powered with high frequency bipolar tailored voltage pulses. It has a configuration as shown in Figure 6k, but water and air are introduced in the reactor with high flow rate of 5 L/min and 100 to 200 L/min, respectively, causing a whirlpool. For this reactor, gas flow rate was shown to have negligible effect on decomposition efficiency of methyl orange.

#### *3.3.4. Arc discharge over water surface*

Gliding arc discharge above a water surface (Figure 7a) is a popular approach for water treatment with plasma. In this reactor type, two diverging electrodes are placed above a water solution. An electric arc forms at the shortest electrode gap and glides along the electrode's axis under influence of a gas flow directed toward the water surface. The arc length increases on moving and its temperature decreases, turning the arc from thermal plasma into quenched plasma while breaking into a plume. A new arc then forms at the narrowest gap and the cycle continues. Unfortunately, many publications on this type of reactor are unclear about the use of AC or DC voltage, but AC power is definitely a common choice. Important research in this field has focused on enlarging the plasma treated water surface with adjustments in design. One possibility is to use a couple of controlled electrodes in between the electrode gap to facilitate breakdown, increasing current intensity and allowing a larger interelectrode distance [136]. Another option is to use three main electrodes supplied by two power sources, as proposed by Burlica et al. [137] (Figure 7b). Both approaches have shown to increase reactor efficiency. Gliding arc discharge can also be used for the treatment of falling water films, as shown by Lesage et al. [123] (Figure 7c). Arc discharge with an active water electrode is less commonly researched (Figure 7d). According to Janca et al. [138], the energy efficiency of such system was found to be strongly dependent on the type of discharge produced, such as arc or gliding arc. For more detailed information on water treatment by means of gliding arc, the reader is referred to the review by Brisset et al. [136].

#### **3.4. Spray discharge reactors**

#### *3.4.1. Low-energy spray discharge reactors*

Although spray discharge reactors have received more attention in recent years due to their high reported energy efficiencies [42–44], still more research is required to characterize and optimize them. One of the most common spray reactors has a wire-to-cylinder geometry

**Figure 7.** Types of gliding arc discharge reactors over water surface: (a) standard configuration, (b) configuration with extra main electrode, (c) gliding arc discharge to falling water film, and (d) gliding arc discharge with active water electrode.

(Figure 8a), similar to wetted-wall reactors. This reactor type is always operating with positive pulsed corona or streamer discharge, according to our literature review. In the study of Kobayashi et al. [139], different spraying nozzles were used to investigate the influence of water location on the energy efficiency of indigo carmine decomposition. As the results showed, spraying water as droplets into the discharge area is more effective than making it flow as a water film on the inner reactor wall. Moreover, droplets that were sprayed close to the reactor wall underwent 1.5 times faster decolorization than droplets near the wire electrode. Energy efficiency was found to be independent of droplet size for same water flow rate [140].

Sugai et al. adjusted this reactor by addition of packed-bed of pellets (Figure 8b) or fluorocar‐ bon wires (Figure 8c) in order to increase the droplet retention time in the discharge space [141]. The packed pellets were hollow polyethylene balls with 14 holes per ball to increase discharge. The fluorocarbon wires were woven as insulation grids in the outer cylindrical electrode. Addition of the pellets decreased the energy efficiency significantly, due to the narrowing of the discharge space. The fluorocarbon wires, however, kept the discharge space unaltered and increase the energy efficiency with 2–10%.

In the study of An et al. [142], a similar electrode configuration was used as in Figure 8a, but with tooth wheels assembled on the inner electrode (Figure 8d). Here, droplets were not created by spraying, but due to condensation of steam under influence of up-flowing air. Alternatively, positive pulsed corona can also be generated in a spray reactor around wire anodes in parallel with two grounded plate cathodes [143], as depicted in Figure 8e. In another approach, water is sprayed in between rod electrodes that are each surrounded with a dielectric barrier (Figure 8f). Monopolar pulsed voltage of both polarities is reported in literature [144, 145]. In the study of Wang et al. [146], grounded water was sprayed from dielectric nozzles in proximity of a high-voltage dielectric barrier plate electrode (Figure 8g). When AC high voltage was applied to the plate, the by electrostatic induction generated electrostatic force pulled the water droplets to the glass dielectric layer. The energy efficiency of indigo carmine decompo‐ sition depended on both voltage amplitude and air gap distance, for which optimal values were found. Another interesting reactor type is based on the electrospray process to simulta‐ neously generate and treat water spray under influence of high voltage, as shown in Figure

Electrical Discharge in Water Treatment Technology for Micropollutant Decomposition http://dx.doi.org/10.5772/61830 453

**Figure 8.** Types of low-energy spray discharge reactors: (a–c) wire-to-cylinder corona or streamer reactor without or with packed-bed of pellets or fluorocarbon wires for increased droplet retention, (d) mist droplets in tooth wheel-tocylinder corona reactor, (e) multi-wire-to-plate corona reactor, (f) spray through DBD rods, (g) spray from grounded electrode to DBD plate, and (h) electrospray reactor.

8h. In the study of Elsawah et al. [147], water is treated that way with positive pulsed corona electrospray.

#### *3.4.2. Spray arc reactors*

Water spray can also be introduced in gliding arc discharge for treatment, as depicted in Figure 9a. This method is found to be more energy efficient than gliding arc discharge over a water surface as discussed in Section 3.3.4 [136, 137]. The decomposition of 4-chlorophenol in such spray reactor became more energy efficient with increasing gas–water mixing rate [148]. Efficiency is higher with electrodes from stainless steel than for aluminum or brass electrodes. Also here, extending the plasma volume by the use of a controlled electrode couple or an extra main electrode enhances energy efficiency (Figure 9b) [137]. The process can also be optimized by introducing the water with a flat spraying nozzle perpendicular to the gas flow to improve the contact with the plasma (Figure 9c) [149]. For more detailed information on water treatment by means of gliding arc spray reactors, the reader is referred to the review by Brisset et al. [136]. Water can be treated as well with a DC plasma torch, where it is usually directly introduced into the torch as plasma forming gas. A discussion on this treatment method is provided by Brisset et al. [44].

**Figure 9.** Types of spray gliding arc reactors: (a) standard configuration, (b) configuration with extra main electrode, and (c) water jet under angle through gliding arc discharge.

#### **3.5. Hybrid reactors**

Electrohydraulic discharge and gas phase discharge can simultaneously be generated when a high-voltage electrode is placed in the water phase with a grounded electrode above the water surface in the gas phase. Mostly, positive pulsed corona is generated in these systems reported in literature. Several electrode configurations are possible, such as an underwater pin to plate in gas [150] (Figure 10a) and an underwater pin to multipin in gas [151] (Figure 10b). In some reactors, a second high-voltage electrode is placed in the gas phase, either powered by the same high-voltage source [151] (Figure 10c) or by a second one [152] (Figure 10d). Also here, energy efficiency can be enhanced by discharge formation in externally applied bubbles, leading to hybrid reactors that combine bubble discharge with gas phase discharge. One example is given by Ren et al. [153], where bubbles are formed on high-voltage nozzle electrodes located underneath a plate electrode in the gas phase (Figure 10e). Hybrid reactors with discharge in both water and gas phase are sometimes proposed for the treatment of gaseous and aqueous pollutants simultaneously, as in the case of volatile pollutants [44, 46]. Their energy efficiency for organic decomposition is moderately higher than the one of hydraulic discharge reactors [42, 43].

Another type of hybrid reactor combines the treatment of a falling water film and droplets. This situation naturally occurs by spraying water from a shower nozzle at an angle. In the study of Kobayashi et al. [139], a wetted-wall hybrid reactor with inner wire anode was used, where water solutions from the falling film and from the droplets were collected separately (Figure 10f). Indigo carmine was decomposed 0.57 times faster in the droplets than in the water

Electrical Discharge in Water Treatment Technology for Micropollutant Decomposition http://dx.doi.org/10.5772/61830 455

**Figure 10.** Types of hybrid reactors: (a, b) underwater high-voltage pin to plate or multipin in gas, (c) reactor from Lukes et al. [151], (d) reactor from Lukes et al. [152], (e) bubble discharge on underwater nozzle electrode to plate in gas, (f) wetted-wall spray corona reactor with inner wire electrode, (g) wetted-wall spray DBD reactor with outer barri‐ er, and (h) electrospray through wire-to-mesh corona reactor.

film, as could be expected from the higher energy efficiency of spray discharge reactors as compared to falling water film reactors. In the study of Nakagawa et al. [154], the water film was formed on the dielectric barrier which separated the inner rod anode and the grounded mesh surrounding the barrier (Figure 10g). For low water flow rate, the energy efficiency of rhodamine B decomposition was similar for 3 of such reactors with different inner diameter and barrier thickness. For higher flow rates, however, one reactor performed significantly better. This reactor had larger inner diameter and equal barrier thickness compared to one of the other reactors.

A special case is found when droplets from an electrospray are treated a second time with plasma discharge. In the study of Njatawidjaja et al. [155], electrostatically atomized droplets passed through pulsed corona discharge in between a wire-to-mesh electrode configuration (Figure 10h). In both parts of the reactor, positive polarity performed better than negative one for the decomposition of Chicago sky blue dye. In the electrospray, positive DC voltage produced a larger number of finer droplets with a wider spray angle than in the case of negative voltage.

#### **3.6. Remote discharge reactors**

The concept of remote discharge reactors for water treatment is not new. An early example is ozonation, where ozone is generated by means of plasma discharge in gas phase and subse‐ quently transported toward the solution under treatment. More recently, electrical discharge reactors have been developed where remotely generated plasma gas is bubbled through the solution. As main difference, the plasma gas does not only contain ozone, but also other reactive species, such as H2O2 and OH. In the study of Tang et al. [156], this is accomplished by using humid air as feed gas of a DBD gas phase reactor (Figure 11a). With corona discharge in dry air, positive and negative ions can be generated in addition to ozone for water treatment, as in the air ionization device reported by Wohlers et al. [157]. Yamatake et al. compared the bubble discharge reactor from Figure 3k with ground electrode in contact with water to a reactor where plasma is generated separately from the water and subsequently bubbled through the solution (Figure 11b) [158]. Both reactors had identical electrode configuration, used oxygen as feed gas, and were powered by the same positive DC voltage source. Acetic acid was not decomposed in the remote discharge reactor, while decomposition was significant in the bubble discharge reactor. This difference was explained with the production of oxygen radicals, which acted as main oxidant during bubble discharge but had too short lifetime to reach the water in the remote discharge reactor.

**Figure 11.** Types of remote discharge reactors: (a, b) plasma gas bubbling reactors, (c–e) plasma gas bubbling reactors with UV irradiation through quartz barrier, (f) underwater DBD plasma jet, (g) DBD plasma jet over water surface, (h) metal strips on quartz disc for DBD above water surface, (i) setup from Dobrynin et al. [159] with removable grounded mesh, (j) reactor from Zhang et al. [160] for production of hydroxyl radical solution, (k) gliding arc reactor from Kim et al. [161] for plasma-activated water production, and (l) reactor with tungsten-triggered microwave discharge electro‐ deless lamp.

Plasma gas bubbling can be combined with UV irradiation by generating plasma inside a quartz tube submerged in the solution. Several variations are reported in literature, where the high-voltage electrode inside the tube is a screw [162], a spiral wire [163], a rod covered by dielectric barrier [164], or a glass tube filled with NaCl solution [165]. A metal mesh surround‐ ing the tube [162] or a metal rod in contact with the water [165] serves as ground electrode (Figures 11c–d). DBD plasma is usually generated inside the tube with AC power. The working gas is pumped in the tube and subsequently bubbled through the solution by means of a gas diffuser or a series of air distribution needles (Figure 11e) [164]. Comparison with other oxidation techniques has indicated that such reactors may be competitive technology to other plasma systems such as the hybrid reactor by Nakagawa et al. [154] (Figure 10g) and to photocatalytic oxidation [165, 166].

A system closely related to these plasma gas bubbling reactors is a submerged DBD plasma jet for water treatment. In the study of Foster et al. [167], a nanosecond pulsed DBD plasma jet is investigated for oxidation of aqueous organic pollutants (Figure 11f). Its decomposition efficiency was higher than the one reported in literature for glow discharge and pulsed corona discharge [167]. The energy efficiency of methylene blue decomposition significantly dropped with increasing treatment time. As the results suggest, using multiple plasma jets powered in parallel can improve the process significantly. The plasma jet can also be placed above the water surface (Figure 11g), as is done for the production of plasma-activated water by Ma et al. [168]. Nevertheless, application of non-thermal plasma jets for water treatment is relatively uncommon for now.

A few systems are reported in literature where plasma is generated above a water surface without direct contact with the solution. In the study of Olszewski et al. [169], copper strips adhered to both sides of a quartz disc were used as electrode configuration for methyl orange decomposition with AC DBD plasma (Figure 11h). In the study of Dobrynin et al. [159], inactivation of spores was investigated by means of the DBD electrode system of Figure 5c above a removable mesh for reference experiments (Figure 11i). Experiments with and without the mesh were compared to reveal the role of UV irradiation.

Recently, a new approach is gaining popularity in which water treated with plasma, often termed plasma-activated water, is added to the solution under treatment. Plasma-activated water contains several long-living oxidants that are able to inactivate biological organisms. According to Zhang et al., long-living aqueous hydroxyl radicals were produced in their setup by mixing DBD-treated humid O2 gas with water (Figure 11j) [160]. This hydroxyl radical solution was sprayed in a sea enclosure to effectively inactivate red tide organisms. Plasmaactivated water can be produced by means of any of the reactors mentioned above, but gliding arc discharge is most commonly used. Figure 11k depicts the gliding arc reactor used by Kim et al. [161] for plasma-activated water generation utilizing a vortex flow with two circular disk electrodes. Introducing the produced water–air mixture in the solution under treatment through microbubble generators significantly enhanced the process. Up to now, plasmaactivated water is used for disinfection of water, while its effects on organic contamination are still largely unknown.

In principle, plasma technology is also frequently used in water treatment technology solely as UV source in plasma lamps or excimer lamps. An upcoming new technology is the micro‐ wave discharge electrodeless lamp, which self-ignites under influence of microwave power. As an example, Figure 11l schematically shows a tungsten-triggered microwave discharge electrodeless lamp proposed by Horikoshi et al. [170] for low microwave power levels. The tungsten wire was embedded in a synthetic quartz tube attached to the lamp system to act as a trigger. In that manner, the high microwave power usually required for self-ignition in aqueous medium is avoided.

### **4. Influence of working parameters on energy efficiency**

In addition to reactor design and materials, there are several other factors that influence reactor energy efficiency, which need to be considered for reactor optimization. Roughly, these factors can be split into two groups: working parameters determining reactor operation and solution parameters. Since solution parameters, such as water temperature, pH, conductivity, and water matrix, are hard to control, especially for large volume of influent water, they are not discussed in this chapter. For a discussion on their influence, the reader is referred to [44]. In contrast, operational parameters such as applied voltage characteristics, working gas, and flow rates of gas and solution are easier to adjust. Therefore, they deserve additional attention for further reactor optimization. In this section, we will shortly review the importance of voltage waveform, working gas, and gas and water flow rates. Their influence on reactor efficiency for organic decomposition will be illustrated with examples from literature.

For a given reactor, energy efficiency depends on voltage-related parameters. In the pulsed bubble discharge reactor of [92] and the positive pulsed streamer discharge in wetted-wall reactor of [129], energy efficiency increased for rising voltage amplitude. In a negative pulsed DBD falling water film reactor [171] and a positive pulsed corona electrospray reactor [147], however, increasing voltage amplitude reduced the energy efficiency. In the gas phase DBD reactor of [169], the interruption period of pulse-modulated AC voltage had no effect on methyl orange degradation, while lowering its duty cycle from 100% to 25% increased energy efficiency 2.11 times. The authors explained the latter effect with additional dye degradation during plasma off time under influence of long living reactive species such as O3 and H2O2. Sinusoidal voltage frequency is an important parameter, as it can lead to different plasma phenomena, which explains the distinction of AC, radio frequency, and microwave discharge. Nevertheless, very limited information is available in literature on the dependence of the energy efficiency of the sinusoidal frequency for a given reactor design. In the study of Lesage et al. [122], no change in energy efficiency was observed for 4-chlorobenzoic acid decomposi‐ tion with voltage frequency increase from 500 Hz to 2000 Hz for AC powered DBD over moving water film. Increasing AC frequency from 1.5 kHz to 15.6 kHz kept the energy efficiency of a coaxial falling water film DBD reactor in the same order of magnitude as well, in spite of the additional heating that resulted from the higher frequency [172]. In the case of pulsed dis‐ charge, pulse properties such as rise time and width are expected to be important. For pulsed positive corona discharge in humid O2/N2 atmosphere, the energy efficiency of radical and excited species production increases with decreasing pulse width [173]. This is in agreement with the higher efficiency of indigo carmine decomposition for shorter pulse width observed by Sugai et al. [174] for a positive pulsed streamer discharge spray reactor. A faster pulse rise rate generated thicker streamers and a higher energy efficiency in the same reactor [175]. Positive pulsed arc electrohydraulic discharge was reported to have increased the energy efficiency of sulfadimethoxine when pulse duration was brought back from 100 µs to 20 µs [51]. Remarkably, also pulse frequency can influence reactor efficiency. In a bubble reactor with positive pulsed corona, the energy efficiency of 2,4-DCP degradation increased with increasing pulse frequency [176]. In contrast, the energy efficiency of indigo carmine reduction dropped for increasing pulse frequency in a positive pulsed streamer wetted-wall reactor [129], in a negative pulsed DBD spray reactor [144], and in a positive pulsed DBD hybrid film and spray reactor [154]. In a negative pulsed DBD falling water film reactor [177] and a positive pulsed corona spray reactor [143], no influence of the frequency was observed. In bipolar pulsed electrohydraulic reactors, breakdown voltage decreases with frequency [62, 96]. Reversing voltage polarity can also cause significant changes in plasma properties and thus treatment efficiency. In the study of Lee et al. [96], positive polarity of pulsed corona in electrohydraulic discharge greatly enhanced the energy efficiency of methyl orange decom‐ position in pin-to-plate electrode configuration in comparison with negative polarity. The authors explain this observation with the space charge effect, which causes positive corona streamers to be faster and longer, hence increasing radical production and plasma–water contact surface. In the study of Yasuoka et al. [178], higher efficiency was observed as well for positive polarity in a DC bubble discharge reactor. For phenol decomposition in gas phase discharge reactors with a 20% O2 and 80% N2 atmosphere, negative DC was found to be more efficient than positive one by Sano et al. [179], while according to Miyazaki et al. [108], no significant difference was seen for both polarities with pulsed power. In a pulsed DBD falling water film reactor, better results were obtained with negative polarity as compared to positive one [171]. In the bubble discharge reactor of Figure 3h with O2, He, Ar, or Ne bubbles, positive DC voltage performed better than negative one for the decomposition of interfacial active agents [83]. For negative polarity, electrolysis occurred with formation of hydrogen and oxygen gases. For positive polarity, the interfacial anion agents were more concentrated at the bubble surface due to electrostatic attraction. Positive plasma species collided with the water surface, where a cathode drop is formed. These unknown species likely enhanced the decom‐ position of the agents.

The working gas determines many plasma features for a given input voltage waveform, such as breakdown voltage, electron density and temperature, plasma homogeneity and intensity, generated reactive species, etc. Air is the most frequently used working gas for water treatment plasma reactors due to its wide availability. Pure oxygen gas, however, is often found to give more efficient organic degradation, while nitrogen gas leads to lower efficiency [42, 44, 50]. This can be partly explained with formation of OH radicals and O3 in oxygen. In N2 containing gases, however, toxic aqueous nitric products are generated, which decrease solution pH and act as scavengers of oxidants such as OH radicals [50]. Noble gases like helium and argon are sometimes used, especially in bubble discharge reactors. Often, argon leads to the faster decomposition of phenols but performs worse than oxygen for other compounds [50]. In a wetted-wall reactor (Figure 6b) with argon, streamer discharge has been observed which had slightly better phenol decomposition energy efficiency than corona in oxygen [128]. Noble gases, however, are expected to be less economically feasible for use on larger scale due to their high price. Interestingly, treatment efficiency of electrohydraulic discharge reactors can also be altered by bubbling different gases through the solution under treatment, where bubbles are kept away from the discharge zone. Sahni and Locke observed a decrease in nitroform anion decomposition by pulsed underwater corona when the solution was oxygenated with O2 gas or deoxygenated with N2, He, or Ar gas as compared to discharge without prior bubbling, but the authors could not explain this effect [180]. Also, water content of the working gas should be taken into account, as it influences formation of important oxidants such as OH and H2O2.

Gas flow rate is another factor that needs to be taken into consideration. In three bubble discharge reactors, increasing bubble flow rate enhanced the decomposition process [158, 181, 182], while according to Reddy et al. [183] no significant effect was observed. In the hybrid reactor with bubbles from Figure 10e, oxidation rate first increased and then reached a stationary value with rising gas flow rate [153]. In the remote discharge reactor with bubbling from Chen et al. [184], increasing gas flow also enhanced decomposition. For gas phase discharge reactors, the influence of gas flow rate seems less pronounced. In a positive pulsed corona-like discharge over water, phenol decomposition was slower with increasing oxygen flow and slightly dropped with increasing airflow, while argon flow rate had no influence [185, 186]. In a pulsed DBD falling water film reactor, the effect of the oxygen flow rate on methylene blue degradation was not significant [171].

In reactors with moving solution, the water flow rate often influences the decomposition process. In coaxial falling water film reactors with corona discharge, phenol degradation rate was unchanged with faster water flow for positive pulsed voltage [128], while it dropped in case of negative DC, which was attributed to higher roughness of the water surface [127]. In a pulsed DBD falling water film reactor, the energy efficiency of methylene blue degradation decreased with increasing water flow rate [171]. In contrast, 4-chlorobenzoic acid was decom‐ posed faster with increasing water flow for AC gliding arc discharge over falling water film [123]. Measurements in spray discharge reactors indicate the existence of an optimal water flow rate. In a positive pulsed streamer spray reactor, the energy efficiency of indigo carmine decomposition initially increased and then stabilized with increasing water flow rate due to saturation of aqueous ozone [140]. Moreover, the efficiency was independent of droplet size. In a similar corona reactor, the energy efficiency of oxalic acid decomposition first increased with rising water flow rate, reached a maximal value and subsequently dropped again [187]. The optimal water flow rate increased with applied pulse frequency. Rising water flow rate decreased breakdown voltage in a positive pulsed DBD spray reactor [145]. For the hybrid positive pulsed DBD spray and falling water film reactor of Figure 10g, the energy efficiency of rhodamine B decomposition enhanced with increasing water flow [154].

### **5. Summary and concluding remarks**

Removal of hazardous micropollutants is often insufficient by means of modern conventional wastewater treatment plants. Preventive measures and optimization of conventional biologi‐ cal treatment are suggested as most cost-effective solutions. Nevertheless, preventive meas‐ ures are limited by increasing demand, while the optimization of conventional techniques often has negligible effect on many persistent micropollutants. Therefore, advanced treatment techniques such as electrochemical separation, activated carbon, nanofiltration, and reverse osmosis have recently received more attention for their effective removal of micropollutants. These techniques, however, are associated with high costs and the additional problem of hazardous concentrate disposal. Advanced oxidation techniques are a promising alternative, as they are the most effective available methods to decompose biorecalcitrant organics. As a main drawback, their energy costs are high up to now, preventing their implementation on large scale. Alternatively, their application can be limited to the treatment of important micropollutant sources, such as hospital and industrial effluent.

Among the advanced oxidation techniques, water treatment by means of electrical discharge takes an interesting place since it is able to generate a wide spectrum of oxidative species, leading to a low selectivity of the degradation process. Further, the optimization of this technology is complex due to the wide variety in reactor design and materials, discharge types, and operational parameters. In this chapter, plasma reactors are comprehensively classified based on their design and materials, in contrast to other reviews where focus lies more on applied voltage and discharge type. Six main reactor types are distinguished. In electrohy‐ draulic discharge reactors and bubble discharge reactors, plasma is generated directly in the liquid bulk, respectively, without and with external application of bubbles. In gas phase discharge reactors and spray discharge reactors, plasma is generated in het gas phase, respectively, over a water bulk or film and in contact with water drops or mist. Reactors that use a combination of these types simultaneously are classified as hybrid reactors. In the last type of reactors, referred to as remote discharge reactors, plasma is not generated in direct contact with the solution under treatment.

Most commonly, electrohydraulic discharge reactors use pulsed arc or positively pulsed corona discharge, where electrode material can have substantial influence on organic decom‐ position after plasma contact due to formation of erosion particles in water. For the case of arc discharge, energy efficiency is reported to be dependent on interelectrode distance. For pulsed corona, high-voltage pin curvature radius is an important parameter as well. The corona plasma volume can be enlarged by replacing the high-voltage pin with a multipin electrode or a high-voltage electrode covered with a thin porous ceramic layer. Diaphragm and capillary discharge are expected to have similar plasma features to corona discharge in pin-to-plate electrode configuration, which is in agreement with the similar energy efficiency. Contact glow discharge electrolysis is another common type of electrohydraulic discharge, where electrode material also plays an important role. In more exotic types of electrohydraulic discharge reactors, plasma formation is preceded by cavitation under application of RF or microwave power.

Adding external bubbles in the water bulk has the advantages of easier plasma onset, imme‐ diate mixing of the solution, minimizing electrode erosion and increasing radical density. Most commonly, bubbles are generated by pumping gas through a dielectric or an electrode, which is shaped as a nozzle, perforated plate, or porous ceramic. Choices in bubble gas, nozzle or perforation material, shape, dimensions, and orientation strongly influence plasma properties, which complicates reactor comparison. Often, the high-voltage electrode is positioned directly in contact with the bubble or in de gas phase in contact with the bubble. In that case, energy efficiency can be enhanced by increasing the number of nozzles or holes. Alternatively, bubbles can be positioned in between submerged electrodes. Some common bubble discharge reactors use a coaxial DBD configuration, where the gas is bubbled in axial direction. Energy efficiency in these systems can be increased by using a double barrier or adding glass beads in the water bulk. More exotic types of bubble discharge reactors have been reported in literature, with promising first results.

Corona and glow discharge in gas phase over grounded water bulk or film is mostly generated with pulsed power. Negative pulsed corona over flowing water film in a pilot system has been shown to have better or comparable energy efficiency than other advanced oxidation proc‐ esses. Based on one report, energy efficiency of organic decomposition seems to be independ‐ ent of the type of discharge, voltage amplitude, polarity of the applied voltage, and amount of pin electrodes. In contrast, DBD over water bulk or film has an energy efficiency that is reported to increase with increasing voltage amplitude and decreasing interelectrode distance. Interelectrode distance can be decreased substantially by adding a porous ceramic segment at the water surface. Movement of the water phase by stirring or by making it flow as a film along the discharge increases energy efficiency substantially, making water flow rate an important parameter for optimization. Based on this principle, coaxial reactors with falling water film are gaining more popularity. Such reactors can be further optimized by adjusting gas flow direction and electrode and barrier dimensions. A last common gas phase discharge reactor uses arc discharge over the water bulk or film. Larger arc-treated water surface and energy efficiency can be reached by using a couple of controlled electrodes or a second high-voltage electrode.

Spray discharge reactors have received more attention in recent years due to their high reported energy efficiencies. In the case of positive pulsed corona, treatment is more efficient for droplets near the inner reactor wall, while droplet size has no influence on energy efficiency. Such reactors can be further optimized by adding fluorocarbon wires along the inner reactor wall for larger droplet retention time. Other spray discharge reactors treat droplets with DBD, electrospray or arc discharge. In the case of gliding arc, the process can be optimized by addition of a couple of controlled electrodes or a second high-voltage electrode and by introducing the water perpendicular to the gas flow to improve contact with plasma. DC plasma torches can also be used for water treatment by injecting the solution into the torch as plasma forming gas.

One type of hybrid reactor is designed by placing a high-voltage electrode in the water phase and a ground electrode or second high-voltage electrode in the gas phase above the water surface, without or with the addition of bubbles. Their energy efficiency for organic decom‐ position is moderately higher than the one of hydraulic discharge reactors. Another hybrid reactor type naturally occurs by spraying water from a shower nozzle at an angle, causing a falling water film in combination with spray. In such case, organic decomposition in droplets is more efficient than in the water film. A more exotic hybrid reactor that deserves more attention consists of a spray discharge reactor where droplets are formed by electrospray with additional plasma treatment afterwards.

Remote discharge reactors can be encountered in many configurations. In the most standard design, plasma gas is remotely generated and sequentially bubbled through the solution, as in the well-known example of ozonation. Plasma gas bubbling can be combined with UV irradiation by generating plasma inside a quartz tube submerged in the solution. Research has indicated that such reactors may be competitive technology to systems with direct plasma treatment. Plasma can also be used solely for UV irradiation, as in plasma lamps, excimer lamps and microwave discharge electrodeless lamps. Recently, the generation and application of plasma-activated water for water disinfection is gaining popularity, while its effects on organic contamination are still largely unknown.

Additionally, the importance of voltage waveform, working gas and flow rates of gas, and water for further optimization are shortly reviewed in this chapter. Energy efficiency has different dependency for different reactor types on voltage parameters, such as voltage type, amplitude, polarity, sinusoidal frequency, pulse rise time, pulse duration, and pulse frequen‐ cy. For example, positive polarity causes higher efficiency of electrohydraulic and bubble discharge reactors in all considered cases, explained with the space charge effect. On the other hand, negative polarity gives better performance in gas phase discharge reactors. The choice of working gas can significantly alter plasma chemistry and therefore treatment efficiency and by-product formation. While atmospheric air is often chosen due to its wide availability, oxygen generally enhances the process. Argon often performs better for phenol degradation but is expected to be less economically feasible for use on large scale. Increasing gas flow rate typically enhances decomposition in bubble discharge and remote discharge reactors, whereas its effect is less pronounced in gas phase discharge reactors. Also, adjustment of water flow can significantly increase energy efficiency. Measurements in spray discharge reactors, for instance, indicate the existence of an optimal water flow rate. While dependency of a few of these parameters on energy efficiency seems adequately established for a limited number of reactor types, a deeper literature study and more experimental investigation are required for additional confirmation and a better understanding in most of the cases. Due to the wide variety in plasma reactors and the distinct, unique features of every reactor, researchers are motivated to report new results in this field, including clear descriptions of reactor design and materials.

One needs to keep a few additional influences in mind when interpreting energy efficiencies reported in literature. First of all, solution parameters such as water temperature, pH, con‐ ductivity, and water matrix can have significant effect on plasma chemistry and reactions in the water bulk. Often, deionized water at room temperature is used as solvent in plasma reactors. In other cases, however, deviations in energy efficiency can be caused by a difference in solution parameters. Also, it should be remarked that micropollutant measurement is generally performed a certain time after plasma treatment with analytical chemistry methods, such as gas chromatography–mass spectrometry and liquid chromatography–mass spectrom‐ etry. During this time, postreactions with long-living oxidants can occur, decomposing the micropollutant to a greater extent. This effect is usually neglected by researchers, complicating accurate comparison between reported energy efficiencies. On the positive side, this aging effect can be beneficial in applications where sufficiently long hydraulic retention time is possible after plasma treatment. Last but not least, most of the reported electrical discharge reactors only treat small water volumes in batch mode without or with recirculation of water. As a result, the determined energy efficiency can be significantly different for the same reactor type in single-pass mode, where water is flowing through the system only once. The latter case is more representative for industrial application and thus deserves more attention.

In this chapter, only improvement of the plasma process in terms of reactor design, materials, and working parameters is discussed. Further optimization can be achieved by combining plasma discharge with other advanced treatment methods, such as adsorption, Fenton's reagent, photocatalysis, and ultrasonication. Such combinations have been reported before but require additional attention and further exploration. As should be noted, plasma technology can also be used for synthesis, pretreatment, regeneration, and posttreatment of materials and matter involved in water treatment processes, such as nanotubes, membranes, activated carbon, excess sludge, and organic concentrate from filtration. In-line application of these methods during the water treatment process needs to be considered as a possible alternative to direct water treatment with advanced oxidation processes since energy demand and overall costs can be pressed significantly this way. However, more experimental investigation and thorough cost analysis is necessary to confirm this claim.

Future application of plasma discharge for water treatment will largely depend on its effec‐ tiveness and energy efficiency as compared to other advanced oxidation processes and treatment methods in general, but additional criteria need to be taken into consideration as well. Sustainability, ease of operation, capital costs, and costs related to maintenance, gas input, and additional energy for pumps will also determine whether a system will be adopted on a large scale. Moreover, an extensive study of generated oxidation by-products and long-living oxidants in treated water is necessary to assure that overall toxicity is consistently and sufficiently decreased after plasma treatment. Up to now, reports on these topics are largely lacking in literature or limited to only a few specific cases.

### **Author details**

Patrick Vanraes\* , Anton Y. Nikiforov and Christophe Leys

\*Address all correspondence to: Patrick.Vanraes@UGent.be

Department of Applied Physics, Research Unit Plasma Technology, Ghent University, Ghent, Belgium

### **References**


### **Study of CO2 Decomposition in Microwave Discharges by Optical Diagnostic Methods** 2

Tiago Silva, Nikolay Britun, Thomas Godfroid and Rony Snyders 1∗

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/60692

#### **Abstract**

The increasing of the carbon dioxide (CO2) release into the atmosphere is undeniably one of the biggest concerns for the twenty-first century. Among the different strategies proposed for reduction of CO2 emission (carbon capture and sequestration (CCS), renewable energies, etc.), low-temperature plasma technology offers an alternative and rather efficient way to convert CO2 into the valuables chemicals (e.g. syngas) which can be stored and used afterwards. Several CO2 decomposition plasma-related approaches have been proposed in the literature, all having a main task: increasing the energy efficiency associated to the decomposition process, while keeping the conversion rate at reasonably high level. This task is especially challenging since many kinetic mechanisms of CO2 decomposition in low-temperature discharges are not yet well-known, such as the vibrational excitation which plays a key role in achieving high decomposition rates. In this chapter our recent research efforts associated with the experimental study of the CO2 decomposition in microwave surfaguide low-temperature discharges are presented. The research was focused on the systematic investigation of the basic plasma parameters. The discharge area of the reactor was characterized by optical emission spectroscopy using the light emitted from spontaneous relaxation of excited species in plasma. The critical parameters such as gas temperature and dissociation rate were evaluated. In addition to this, the post-discharge area was characterized by two-photon absorption laser-induced fluorescence and gas chromatography techniques in order to investigate the exhaust gas composition. All together, the results overviewed in this chapter provide interesting insights into different kinetic mechanisms of CO2-containing discharges, which play an important role in the CO2 decomposition process.

**Keywords**: CO2 decomposition, microwave discharge, optical emission spectroscopy, two-photon absorption laser-induced fluorescence, gas chromatography

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

### **1. Introduction**

The combustion of coal, oil, and natural gas supplies approximately 87% of the primary energy used worldwide [1]. These anthropogenic activities are the main sources of CO2 emission, which is well known for its contribution to the greenhouse effect and planetary heating. Currently, we are burdening the atmosphere of our world with an additional 40 billion tons of CO2 every year [2]. Furthermore, the rate of these emissions is dramatically increasing as the world population grows and new economies are developed. As recently reported by the National Oceanic and Atmospheric Administration (NOAA), the level of CO2 accumulated in the atmosphere has reached a symbolic milestone of 400 parts per million (ppm) [3]. This is the highest level of atmospheric CO2 in human history. In fact, the planet has not experienced these levels of CO2 since the *Pliocene* period (more than three million years ago). At this point, we are facing a global warming trend which is undoubtedly one of the biggest concerns for the twenty-first century. The scientific evidence concerning numerous drastic effects on the actual earth's physical and biological systems is overwhelming. Serious consequences include the melting of glaciers, loss of biodiversity, change in the agricultural productivity, etc. Unless we find efficient solutions to reduce our reliance on fossil fuels, the planet will continue to warm up in the coming decades and greater will be the impact of climate change.

### **1.1. The recycling of CO**<sup>2</sup>

One of the most promising solutions to stabilize the amount of greenhouse pollution in the atmosphere is the so-called *carbon capture and utilization* (CCU) [4]. Such an approach attempts to use CO2, either directly or after chemical conversion, as part of a reaction chain to produce value-added products. The supply of CO2 at local level would be guaranteed in the long term by various sources, from small smoke stacks to large coal-fired plants. In this regard, CCU is a very attractive strategy from an economic point of view, considering that in the near future, large amounts of CO2 will become available as feedstock of nearly zero or even negative cost [5]. In addition, given the possibility of having cheap renewable energy (e.g., wind power during the night period [5]), there is a huge potential in combining the CCU with CO2-free electricity. This last point is considered to be a key aspect toward a more general goal of resource since it combines the efficient reuse of a waste followed by the reduction of fossil fuel combustion [6]. An overview on different ways to convert CO2 with the specific reference of introducing renewable energy in the chemical production chain is given in [6, 7]. Among the different CCU routes, the CO2 hydrogenation is widely discussed nowadays. This strategy is based on the water-gas shift reaction (WGSR):

$$\text{CO} + \text{H}\_2\text{O} \Leftrightarrow \text{CO}\_2 + \text{H}\_2\tag{1}$$

to produce syngas (mixture of CO/H2) which can be further utilized as a building block of methanol, dimethyl ether, liquid hydrocarbons, and other useful compounds via the Fischer-Tropsch process [7]. The intrinsically high energy density of these fuels and their good transportability make them highly desirable. Figure 1 shows an idealized energy cycle that combines renewable energies with CCU valorization through the WGSR. The CO2 is recovered from an emission source and transformed into liquid fuels, provided that a sustainable production of CO via renewable electricity is obtained. This is an extremely practical case in which the net production of CO2 would be zero, while the renewable energy could be stored in CO2-neutral fuels to be integrated in the existing transport infrastructure.

2

**Figure 1.** Generic and idealized energy cycle using captured CO2 and H2O to yield value-added products via renewable electricity.

Another important strategy that can be used to produce syngas is the dry reforming of CO2 with methane. This alternative is also particularly attractive since it converts two of the principal gases responsible for the greenhouse effect. In spite of many advantages related to these approaches, there are, however, some critical issues that need to be addressed. In particular, it has to be noted that most of the CCU routes are associated to highly endothermic reactions (e.g., CO2 conversion in Figure 1). Under classical industrial conditions (e.g., using a typical reactor configuration with packed bed tubes inside a furnace), these reactions are sustained and limited by the rate of heat transfer. This leads to high production costs, which turns the whole conversion chain economically unreasonable [8]. In order to overcome this problem, there is one technology worth investigating: plasma-assisted decomposition. Plasmas provide an ideal environment for CO2 conversion due to the formation of high energetic charged species that can initiate chemical reactions difficult or impossible to obtain using ordinary thermal mechanisms. In particular, plasma electrons can lead to the formation of vibrationally excited molecules, which are able to dissociate through the vibrational ladder-climbing process [9]. Among different types of plasmas that can be used for CO2 decomposition, the so-called cold (also known as nonthermal) plasmas are the most promising candidates. These electrical discharges are characterized by nonequilibrium conditions, under which electrons, ions, and neutral species have different translational and – in case of molecules – internal (ro-vibrational) energies. This results in formation of specific nonequilibrium gaseous media in which endothermic processes with increased energy efficiencies and dissociation rates can be achieved [9].

### **1.2. The CO**<sup>2</sup> **dissociation in cold plasma**

From a practical point of view, all the research involving CO2 conversion in cold plasmas has one common task: increasing the energy efficiency associated to the decomposition process while keeping the dissociation rate at a reasonably high level. For the sake of comparison between different experiments, the energy efficiency rate (*η*) of the CO2 decomposition is usually defined as the ratio of the CO2 dissociation enthalpy (∆*H*) and the injected energy per CO molecule produced in the plasma (*E*CO):

$$
\eta = \Delta H / E\_{\text{CO}}.\tag{2}
$$

Note that ∆*H* depends on reaction mechanism. The most desirable chemical route to produce CO from CO2 would be a process or a sequence of processes requiring the smallest amount of energy. In the most direct way, the CO2 decomposition occurs via:

$$\text{CO}\_2 \rightarrow \text{CO} + \text{O}, \quad \Delta H = 5.5 \text{ eV/molecule.} \tag{3}$$

The atomic oxygen created is able then to react either with another oxygen atoms to form O2 or with vibrationally excited CO2 (denoted as CO∗ <sup>2</sup>) according to [9]:

$$\text{CO} + \text{CO}\_2^\* \rightarrow \text{CO} + \text{O}\_{2'} \quad \Delta H = 0.3 \text{ eV/molecule.} \tag{4}$$

Combining the above equations, the net reaction for the total dissociation process of one CO2 molecule yields

$$\text{CO}\_2 \rightarrow \text{CO} + 1/2\text{O}\_{2'} \quad \Delta H = 2.9 \text{ eV/molecule.} \tag{5}$$

The energy efficiency rate given by Eq. 2 can be rewritten for the total CO2 decomposition *η* (%) according to

$$
\eta = \frac{\chi \cdot 2.9}{E\_m} \text{.} \tag{6}
$$

where *χ* (%) is the CO2 dissociation rate, while *Em* is the specific energy input (in eV/molecule and defined as the ratio of the discharge power over the gas flow rate through the discharge). It is important to stress that in practice, there are many chemical and system inefficiencies (e.g., reverse reactions forming back CO2) that contribute to a non-efficient decomposition. In this context, the ideal scenario is to find the type of discharge with optimal plasma properties (pressure, flow rate, power, etc.) that minimize these unwanted mechanisms. The most common types of plasma discharges used for CO2 conversion include the radio frequency (RF) [10], dielectric barrier discharge (DBD) [11–15], gliding arc plasmatron (GAP) [16, 17], glow discharge (GD) [18], and microwave (MW) [9, 19, 20] (see Table 1).

It is interesting to note that among different plasma sources used nowadays, DBD is probably the most widely used in CO2 conversion research. Although the efficiencies obtained with this source are typically low, the possibility to work at atmospheric pressure under nonequilibrium conditions is very promising. Combined with a catalytic material, these discharges should also improve the selective production of the targeted compounds [15]. In addition, the DBD has a very simple design, which is beneficial for upscaling in industrial applications.



2

**Table 1.** List of various plasma sources (excited at frequency *fd*) used to decompose CO2 and the corresponding energy efficiencies. † refers to this work

On the other hand, the GAPs have also been receiving a special attention in environmental applications because they can operate at atmospheric pressures while reaching relatively high efficiencies [21]. Nunnally et al. [17] have recently reported a maximum energy efficiency of 43% on CO2 decomposition. This result was attributed to the reverse vortex flow configuration (also known as *tornado* effect) which increases the residence time of the plasma species and leads to a more uniform gas treatment.

Finally, the MW plasma sources are the ones that (until today) were able to provide the highest efficiencies (∼85%) for CO2 conversion (see Table 1). As pointed out by Fridman [9], the ability to create a strong nonequilibrium environment in MWs leads to the formation of vibrationally excited CO2 states via vibrational-vibrational (VV) exchange reactions, which favors efficient dissociation. It has to be noted, however, that high energy efficiencies in MWs are usually achieved at reduced pressures (∼100–200 Torr), which is not desirable for industrial purposes. Furthermore, MWs operating in the atmospheric regime seem to lead to a clear reduction in the energy efficiency (∼20%) as recently reported by Spencer et al. [19]. This is due to the fact that the degree of nonequilibrium tends to decrease at higher pressure in these discharges. Nevertheless, these considerations make the abovementioned discharges particularly attractive to investigate and to further develop toward an industrial implementation.

#### **1.3. The research strategy used**

The current study is based on the project P07/34 "Plasma-Surface Interaction" of the Interuniversity Attraction Poles (IAP) action supported by the Belgian Science Policy Office (BELSPO), which (among other subjects) combines experimental and theoretical activities related to CO2 decomposition. The research was focused on the systematic investigation of the basic plasma parameters in microwave discharges aiming at enhancing our understanding on the CO2 decomposition process. The characterization of the discharge has been done through different diagnostic methods, namely, optical emission spectroscopy (OES), two-photon absorption laser-induced fluorescence (TALIF), and gas chromatography (GC). Due to their nonintrusiveness, these techniques are particularly useful to characterize the CO2 decomposition in the microwave discharge domain. In this context, we explore the mentioned diagnostic techniques either directly or via development of the additional easy-to-handle tools which, at the same time, may be particularly valuable for maximization of the energy efficiency in real cold plasma discharges targeted to the CO2 conversion. Among different important parameters investigated, a special attention was given to the measurement of the (i) space-resolved CO2 dissociation rate, (ii) gas temperature, and (iii) density of products at the end of the reactor exhaust (post-discharge).

### **2. The experimental part**

The plasma investigated in this work was created and sustained through microwave surfaguide-type wave discharges (here abbreviated as MSGDs). This particular kind of microwave plasma source was already proven to be very efficient to produce atomic species in the discharge volume [22]. Working in the pulsed regime, the MSGDs ensure even more efficient molecular dissociation of diatomic or multi-atomic species [23]. The schematic representation of the experimental setup is shown in Figure 2. Two MSGDs excited at different frequencies of 2.45 and 0.915 GHz have been involved in this work. In either case, the discharge was sustained inside a quartz tube (inner radius *R* = 7 mm and length *L* = 30 cm) surrounded by another (polycarbonate) tube for cooling purposes. The mentioned cooling has been performed using an oil flow of 10 ◦C. The gas mixture was injected from the top of the system and regulated by the electronic mass flow controllers. The pressure range studied was varied from 1 to 90 Torr. The core system was always surrounded by a grounded aluminum grid to prevent a leak of the microwave radiation into the outer space (not shown in Figure 2). The power supply was able to create discrete pulses, typically with lengths in the order of milliseconds to microseconds. A more detailed description of the MSGD can be found in [23]. In respect to the plasma characterization, the OES measurements were performed along the discharge axis, while the TALIF and GC methods were implemented in the post-discharge of the reactor (see Figure 2). The particularities of each diagnostic along with the investigated parameters are described below.

#### **2.1. OES implementation**

OES represents a powerful yet nonintrusive characterization technique based on the measurement and analysis of the light emitted from spontaneous relaxation of excited species in plasma. In the current work, the OES was implemented in the 2.45 GHz MSGD system through an Andor Shamrock-750 spectrometer and an Andor iStarDH740-18F-03 ICCD camera with a built-in digital delay generator. In a first approximation (considering an optically thin plasma), the observed line intensity *I*(*p*, *q*) of a particular electronic transition *n*(*p*) → *n*(*q*) is defined as (see, e.g., [24])

$$I(p,q) = h\nu\_{pq}(4\pi)^{-1}n(p)A(p,q)L,\tag{7}$$

where *hνpq* is the energy gap between the upper and lower level *p* and *q*, *n*(*p*) the density of the emitting level, *A*(*p*, *q*) the radiative transition probability, and *L* the line segment of the plasma along which radiation is collected. A typical emission spectrum recorded at the top of the discharge (*Z* = 7 cm in Figure 2) in a pure CO2 plasma is shown in Figure 3. As one

2

**Figure 2.** Schematic representation of the MSGD together with the diagnostics techniques used for the plasma characterization.

can see, the Angstrom system (*B*1<sup>Σ</sup> <sup>→</sup> *<sup>A</sup>*1Π) of CO occupies the larger part of the spectrum (∼400–700 nm), neighboring the CO third positive system (*b*3Σ<sup>+</sup> <sup>→</sup> *<sup>a</sup>*3Π*r*) [25]. Other CO bands (e.g., triplet, Herman, and Asundi [25]) were not identified during the measurements, and their contribution is accepted to be negligible for further analysis. The presence of the CO<sup>+</sup> <sup>2</sup> ultraviolet doublet (*B*2Σ<sup>+</sup> <sup>→</sup> *<sup>X</sup>*2Π) emission at 289 nm and the atomic O peaks in the near-infrared (NIR) part of the spectrum were identified as well. The continuum part of the spectrum induced by CO–O recombination with a maximum at ∼450 nm is also visible. It is interesting to note that neither C2 nor C emission has been observed in this work, which suggests a negligible amount of atomic carbon produced.

**Figure 3.** Appearance of the emission spectrum of a pure CO2 microwave discharge acquired at *Z* = 7 cm. The notation (*ν* -*ν*) refers to the vibrational transition, where *ν* () stands for upper (lower) vibrational energy level. Reproduced with permission from Ref. [20]. Copyright 2014 IOP Publishing.

The parameters investigated by OES in the plasma phase were the (i) rotational temperature (*Trot*) and (ii) space-resolved CO2 dissociation rate *χ*(*Z*). OES can be used to determine *Trot*, provided there is access to a suitable rotational emission band (*Irot*). In case of Boltzmann equilibrium among the rotational levels, the *Trot* can be determined via [26]

$$I\_{rot} = \frac{\mathcal{C}}{Q\_R} \frac{1}{\lambda^4} \mathbf{S}\_J \cdot \text{Exp} \left( -\frac{F(J')\hbar c}{k\_B T\_{rot}} \right) \tag{8}$$

where *C* is a constant combining all the terms nondependent on the rotational number *J*, *QR* the rotational state sum, *λ* the wavelength of the emitted radiation, *SJ* the *J*-dependent dimensionless Honl-London factors, *F*(*J* ) the rotational energy term (*J* () stands for upper (lower) rotational level), *kB* the Boltzmann constant, and *c* the speed of light. Based on Eq. 8, the so-called Boltzmann plot in the coordinates (*F*(*J* ), Log(*Irot*/*SJ*)) is often applied to extract the *Trot*. This parameter is particularly relevant in CO2 plasmas since it gives a first estimation of the gas temperature *Tgas* [20]. In the current work, the rotational spectra of CO and N2 have been used for rotational analysis. The Honl-London factors of these molecules can be found in [26].

On the other hand, the estimation of the *χ*(*Z*) cannot be directly obtained from the emission intensity since the number densities of the decomposed species are represented by their non-radiative ground states. The simplest way to overcome this issue by OES is through the so-called *corona* model [27] and the actinometry technique (proposed by Coburn and Chen [28]). Briefly speaking, the corona model assumes that upward transitions (forming excited species) are only due to electron collisions from the ground state, while downward transitions occur via radiative decay. Under these assumptions, the creation-loss balance for the upper level *p* can be written according to

$$m\_{\mathcal{S}} n\_{\mathfrak{e}} k(\mathfrak{g}, p) = n(p) \sum\_{\mathfrak{q}} A(p, q) \tag{9}$$

where *ng* is the ground state density, *ne* the electron density, *k*(*g*, *p*) the rate coefficient of excitation from the ground level, and ∑*<sup>q</sup> A*(*p*, *q*) the sum of all radiative de-excitation processes with origin in the *p* level. In order to get quantitative information about the prospected *ng*, the actinometry technique is applied. In this procedure, a small amount of gas (actinometer) with known concentration *nact <sup>g</sup>* and line intensity *I*(*i*, *j*) with transition *n*(*i*) → *n*(*j*) is added into the discharge. From the ratio of both emission intensities *I*(*p*, *q*) and *I*(*i*, *j*) and combination between Eqs. 7 and 9, the following expression can be deduced:

$$m\_{\mathcal{S}} = n\_{\mathcal{S}}^{act} \frac{I(p,q)\upsilon\_{\bar{i}\bar{j}}A(\bar{i},\bar{j})k(\mathcal{g},\bar{i})\sum\_{\eta}A(p,q)}{I(\bar{i},\bar{j})\upsilon\_{pq}A(p,q)k(\mathcal{g},p)\sum\_{\bar{j}}A(\bar{i},\bar{j})}. \tag{10}$$

In the current research, the actinometry was used to measure the density of CO in the discharge, which leads to the *χ*(*Z*) provided that the initial *CO*<sup>2</sup> density is known. A small amount (5%) of N2 was chosen as actinometer due to the proximity of the N2(*C*3Π*u*) (11.05 eV) and CO(*B*1Σ+) (10.78 eV) excited states. Due to this proximity, the populations of these two energetic states should correlate as a result of the electron excitation. The emission lines chosen for the actinometry analysis correspond to the transitions CO(*B*1Σ+)(*<sup>ν</sup>* <sup>=</sup> <sup>0</sup>) <sup>→</sup> CO(*A*1Π)(*<sup>ν</sup>* = 1) and N2(*C*3Π*u*)(*<sup>ν</sup>* <sup>=</sup> <sup>0</sup>) <sup>→</sup> N2(*B*3Π*g*)(*<sup>ν</sup>* = 2), from the Angstrom system of CO and second positive system (SPS) of N2, respectively. A negligible dissociation rate of N2 was assumed. This last point is supported by (i) OES measurements in CO2-5%N2 gas mixtures showing a negligible emission of atomic nitrogen [20] and (ii) CO2-N2 modeling data showing a negligible N2 dissociation at low N2 admixtures [29]. The rate coefficients necessary in Eq. 10 were calculated via the Maxwellian distribution (assuming electron temperature *Te* ∼1 eV as measured in [20]) and the excitation cross sections of CO [30] and N2 [31]. Finally, it is important to mention that Eq. 10 is only valid in CO2 low-pressure discharges with negligible excitation out of metastable levels (e.g., CO(*a*3Π*r*) and N2(*A*3Σ<sup>+</sup> *<sup>u</sup>* )). An extended collisional-radiative model would be required in the high-pressure regime, since the excitation out of these intermediate levels is expected to play an important role as well (see, e.g., [32]).

### **2.2. TALIF implementation**

2

Laser-based plasma diagnostic techniques have largely contributed to understanding plasma kinetics since the 1970s with the development of solid-state lasers [33]. One of these techniques is the TALIF, which is based on the measurement of the excited state fluorescence radiation induced by the excitation of the ground state via resonant absorption of laser photons. The fluorescence signal is proportional to the laser intensity thus allowing detection of the ground state population density. The use of two laser photons is required for most light atoms because the energy gap between the lower and the first electronic excited level exceeds the laser energy produced by conventional dye lasers.

In the current work, the TALIF was applied in the post-discharge region of the 2.45 GHz system through a YAG:Nd pumping laser (Spectra Physics INDI YAG) coupled with a Sirah Cobra Stretch dye laser. The TALIF signal was acquired perpendicularly to the laser beam through an ICCD detector coupled with Nikkor 50 mm lens (see Figure 2). The parameter investigated by TALIF was the oxygen ground state O(3*P*2) density. The excitation of this level was performed via 2 × 225.586 nm laser photons, while the fluorescence was measured through the radiative decay of the O(3S) level at 844.68 nm. An illustration of the actual fluorescence signal is shown in Figure 4. It is important to stress that the data obtained here by TALIF only yields relative densities. In order to achieve absolute results, the induced fluorescence needs to be calibrated (normally via noble gases [34]). Even though such absolute measurements are particularly promising to improve our understanding on the role played by the oxygen atoms in CO2 discharges, they were not covered by the current study.

### **2.3. GC implementation**

GC offers a sensitive detection of stable reaction products at the exhaust of the plasma reactor. These products are separated inside the gas chromatograph due to their distribution between two non-mixable phases: a stationary phase (solid and/or liquid, filled in the so-called separation column) and the mobile phase (carrier gas, flowing through the column and containing the gaseous sample mixture) [35]. Nowadays, the GC is often applied in many kinds of CO2-containing discharges, including CO2–N2 [29], CO2–H2O [36], CO2–CH4 [37], CO2–Ar [38], etc.

In the current work, the gas chromatograph (Bruker) equipped with a carbon molecular sieve column and a molecular sieve 5A column in series connected with a thermal conductivity detector was implemented in the post-discharge of the 0.91 GHz system. As this plasma

**Figure 4.** Schematic of the post-discharge region (above), along with the actual O(3*P*2) image kept by the ICCD detector (below). Dashed rectangle represents the TALIF signal integration area.

source worked in reduced pressure regime, a sampling system was developed between the post-discharge and the gas chromatograph. The low-pressure sample coming from the post-discharge is diluted with neutral gas, in order to reach atmospheric pressure, prior to its injection in the chromatograph. Argon was used as a carrier gas. The CO2 dissociation rate is calculated by

$$\chi = \frac{A\_{\rm CO\_2}^{off} - A\_{\rm CO\_2}^{on}}{A\_{\rm CO\_2}^{off}},\tag{11}$$

where *Aon* CO2 (*Aoff* CO2 ) is the chromatograph signal of CO2 when the plasma is switched on (off). It is important to emphasize that unlike OES, the GC does not provide *in situ* data. However, GC is advantageous over the actinometry method above-described since its applicability is not dependent on the discharge pressure and the accuracy of any population model.

#### **3. The characterization of the discharge area**

In this section, we will discuss the results related to investigation of the pulsed CO2-containing discharges by OES. The energy delivered in the plasma pulse (E*p*) during these measurements was varied from 0.8 J to about 1.2 J. The discharge pressure was kept below 5 Torr in this part of work.

#### **3.1. Spatial analysis (OES)**

The first insight that one can take regarding the plasma composition is based on the discharge color. For instance, as reported by Timmermans et al. [39], the emission of C2 molecules in CO2 discharges (only detected at high pressures due to the significant production of atomic carbon) is represented by a typical green emission. On the contrary, in our case, the light emission from the discharge always showed a strong blue emission due to the presence of CO (see Figure 5). Furthermore, an increase of this light (readily observed by eye), after passing the waveguide (bottom part in this case), has motivated the space-resolved measurements along the discharge tube. Being represented at two different positions in the discharge (top and bottom), the emission spectrum shows certain changes reflecting the molecular decomposition process happening along the gas flow direction (see Figure 5). In addition, one can clearly observe a definite increase of CO Angstrom emission intensity from the top to the bottom of the tube.

2

**Figure 5.** Emission spectra of the CO2 MSGD taken at two different discharge positions (shown by the white arrows). Pressure ∼3 Torr.

The abovementioned spatial increase of CO emission is also visible via time-resolved OES imaging (see Figure 6). These measurements provide a density mapping of the excited species in plasma. A filter that passes wavelengths within the 480 nm range was used as indicator of the CO Angstrom emission. Using a plasma on-time of 600 µs, one can clearly see that the CO emission extends axially along the quartz tube with clear predominance at the bottom region of the discharge.

**Figure 6.** Time-resolved two-dimensional OES imaging of the CO2 MSGD using a 480 nm optical band-pass filter. The iCCD camera was positioned ∼15 cm away from the surfaguide. The mean power applied was 300W with plasma on-time of 600 µs and 50% duty cycle. Each frame is a single-shot picture with a gate step of 20 µs.

In order to clarify the previous observations, the space-resolved temperature measurements were performed. Figure 7 shows *Tgas* calculated via the Boltzmann plot (as described in Section 2.1) through the CO and N2 rotational bands. The given error bars correspond to the statistical errors of the Boltzmann fitting in each case. Note that the space interval Z ≈ 8–13 cm was inaccessible for optical measurements due to the presence of the waveguide. These measurements suggest a linear increase of *Tgas* along the gas flow direction. The results can be interpreted as a trace of the CO2 decomposition process. Indeed, if the CO2 molecules in plasma undergo dissociation along the discharge tube, the collision rate increases toward the gas flow direction, which may explain the observed temperature increase. A possible enhancement of the vibrational energy exchange in the plasma phase may also induce vibrational-translational (VT) exchange, which leads to an increase of the gas heating.

**Figure 7.** *Tgas* determined based on the N2 and CO rotational temperatures at different positions of the discharge tube. The CO2 + 5%N2 gas mixture is used with pulse (period) of 1.0 (1.5) ms. Adapted with permission from Ref. [20]. Copyright 2014 IOP Publishing.

#### **3.2. CO**<sup>2</sup> **dissociation rate (actinometry)**

The investigation related to the CO2 dissociation rate obtained by the actinometry method (described in Section 2.1) is presented in this section. The CO2 + 5%N2 gas mixture with pulse (period) of 1.0 (1.5) ms was used. The actinometry ratio between CO and N2 was acquired 0.7 ms after the beginning of the plasma pulse in order to ensure a steady-state plasma regime. As a result of our measurements, a nonuniform *χ*(*Z*) along the gas flow direction was found, as illustrated in Figure 8. The experimental points of *χ*(*Z*) at the top of the discharge are well described by a linear fit (see Figure 8). However, the strong increase (almost four times) between the extremities of the discharge suggests a fast evolution of *χ*(*Z*) in the waveguide vicinity. To fit the obtained *χ*(Z) data, the so-called logistic growth has been proposed:

$$\chi(Z) = \chi\_0 + \frac{\chi\_{\text{max}}}{1 + e^{-r(Z - Z\_c)}},\tag{12}$$

where *χ*<sup>0</sup> (%), *Zc* (cm), and *χmax* (%) are the initial values of the dissociation rate, the middle position of the discharge, and the maximum value of dissociation rate (defined as an average value at the end of the tube). The *r* (cm−1) is the free fit parameter (see Figure 8). A symmetrical power distribution in the discharge is assumed in this case: *χ* (*Z*=*Zc*) ∼ *χmax*/2. A derivative of Eq. 12 should correspond to the spatial distribution of the power absorbed in the plasma bulk *Pabs* which is used for CO2 decomposition. Such a distribution can be characterized by the correspondent width (FWHM), i.e., a spatial region where the decomposition is most efficient, as shown in Figure 8, where the red curve represents the derivative of the black one. As one can see, the initial increase of *Pabs* coincides with a small experimental increase of *χ*(*Z*) at the beginning of the discharge (*Z* = 4–7 cm), followed by the fast growth of *χ*(*Z*) in the waveguide region where P*abs* is supposed to be maximum.

**Figure 8.** Space-resolved values of the dissociation rate *χ* (left scale) and the power absorbed in the plasma bulk *Pabs* (right scale) measured at *Em* = 23 eV/molecule. Reproduced with permission from Ref. [20]. Copyright 2014 IOP Publishing.

Furthermore, it was verified that *χ*(*Z*) increases with *Em* as shown in Figure 9. It is interesting to note that our measurements show that *χ*(*Z*) likely reaches a steady state at the bottom of the discharge, whereas a definite linear increase in *χ*(*Z*) is always recognized at its top, i.e., before the gas passes the waveguide. Such differences can be associated with the different chemical processes between these two discharge regions, which should be strongly related to the different power absorption channels. It would be interesting to compare these results with those obtained in an extended discharge tube in order to further investigate this last point.

A linear proportionality between the *χmax* and *Em* is also observed in our case, as shown in Figure 10. A similar behavior of the CO2 decomposition rate which grows linearly when increasing the power and/or decreasing the flow rate is found in [40]. At the same time, based on Eq. 6, *η* is found to be ∼6% at different *Em* (see Figure 10).

### **3.3. Gas temperature analysis (line ratio)**

2

The study of *Tgas* via the CO rotational band was further investigated in order to (i) validate the Boltzmann plot results previously obtained and (ii) search for a simple gas temperature formula based on CO spectral synthesis. Figure 11 shows an example of the so-calculated CO(*B*1Σ)(*<sup>v</sup>* <sup>=</sup> <sup>0</sup>) <sup>→</sup> *CO*(*A*1Π)(*<sup>v</sup>* <sup>=</sup> <sup>1</sup>) transition based on Eq. 8. The final theoretical spectrum includes the sum of the different rotational branches (*P*, *Q*, *R* corresponding to ∆*J* = *J* − *J* = −1, 0, +1). To reflect the broadening of the spectral lines, a pseudo-Voigt distribution [41], i.e., a combination of a Gaussian and a Lorentzian function, is used:

**Figure 9.** Space-resolved measurements of the CO2 dissociation rate *χ* for two different *Em* values. The region between about 8 and 14 cm was not optically accessible. Reproduced with permission from Ref. [20]. Copyright 2014 IOP Publishing.

**Figure 10.** Maximum CO2 dissociation rate *χmax* (left scale) along with its energetic efficiency *ηmax* (right scale) as a function of *Em*. Reproduced with permission from Ref. [20]. Copyright 2014 IOP Publishing.

$$\log(p\_\prime w\_G, w\_L) = p \cdot \exp[-4\text{Log}(2)(\frac{\lambda - \lambda\_0}{w\_G})^2] + (1 - p)\frac{w\_L}{w\_L^2 + 4(\lambda - \lambda\_0)^2},\tag{13}$$

where *p* is the contribution of the Gaussian function, *λ*<sup>0</sup> the line central wavelength, and *wG* and *wL* the full width at half-maximum (FWHM) for the Gaussian and Lorentzian profiles, respectively. The Voigt profile parameters were experimentally determined by measuring the shape of the 435.8 nm Hg line (3*P*1–3*S*1) from an Ar-Hg lamp (see Figure 11). The Hg line chosen for determination of the monochromator apparatus function is assumed to have (i) a spectral response similar to the one of the CO spectra and (ii) Doppler, Stark, and van der Waals broadening much smaller than the instrumental broadening [42–44]. The rotational peaks located at *P*<sup>1</sup> = 481.61 and *P*<sup>2</sup> = 482.48 nm were chosen in order to build a line-ratio formula. In this procedure, special attention was paid to find isolated peaks with good visibility and good sensitivity to the gas temperature, as discussed in [45]. To avoid bulky calculations, the expression for *Tgas* is then derived, taking into account the contribution of all

Study of CO2 Decomposition in Microwave Discharges by Optical Diagnostic Methods http://dx.doi.org/10.5772/60692 493

2

**Figure 11.** Synthetic spectrum of the molecular transition from the CO Angstrom rotational band calculated at *Trot* = 470 K. Inset: measured Hg (435.8 nm) emission line (squares) and the corresponding fit (solid line) using a pseudo-Voigt function given by Eq. 13. Reproduced with permission from Ref. [41]. Copyright 2014 OAS Publishing.

the *P*, *Q*, and *R* rotational branches based on a semi-analytical approach, namely, when the peak ratio is determined as a function of *Tgas* for each rotational spectrum synthesized. After linearization in the coordinates (*Log*(*R*), 1/*Tgas*), the final result for the mentioned peaks can be presented in the form:

$$T\_{\mathcal{S}^{\rm as}}(K) = 343(0.36 + \mathcal{L}\log(R))^{-1},\tag{14}$$

where *R* = *P*1/*P*<sup>2</sup> is the peak ratio. The graphical representation of Eq. 14 between 200 and 10 000 K (typical range of cold plasmas) is given in Figure 12. The relative error of the *Tgas* determination relatively to the accuracy of the line intensity ratio (*δTgas*) can be determined after differentiating Eq. 14. As a result, we obtain

**Figure 12.** *Tgas* determination chart based on the line ratio between two spectral peaks in the CO Angstrom band (left scale). The red circles and the solid line represent the calculated points and the linear fit, respectively. The dashed line (right scale) indicates the relative error for the gas temperature – *δTgas*. Reproduced with permission from Ref. [41]. Copyright 2014 OAS Publishing.

$$
\delta T\_{\mathcal{S}as}(\mathcal{K}) = \delta \mathcal{R} \cdot (0.36 + \mathcal{L}og(\mathcal{R}))^{-1}.\tag{15}
$$

The quantity *δTgas* is given as a function of *Log*(*R*) in Figure 12 where the *R* relative error *δR* = 0.05 is assumed. As we can see, the *Tgas* relative error increases dramatically, when *Log*(*R*) *<* 0, exceeding 30% for the temperatures above ∼3000 K. This fact limits the applicability of the proposed method at high gas temperatures.

In order to test the proposed method, the time-resolved *Tgas* was measured in pure CO2 MSGD with a plasma pulse (period) of 1.0 (1.5) ms. The proposed line-ratio method is compared with the results obtained by the Boltzmann plot method and direct comparison between the measured and calculated spectra. Figure 13 shows a reasonable agreement in terms of *Tgas* evolution for these different approaches. The uncertainty (about 11% in saturation) of the obtained *Tgas* by the line-ratio method is still in the range of the errors described by Eq. 15 at *Tgas* ∼ 800 K. These time-resolved measurements also show that, at a fixed *Ep*, the *Tgas* grows at the beginning of the pulse and saturates after about 0.4 ms. These effects were observed previously for the other gas mixtures in the same type of discharges [46]. First effect corresponds to the gas heating at the beginning of the plasma pulse, whereas the second one reflects the equilibrium between the gas heating and heat dissipation and may be related to VT relaxation processes, as suggested in [46].

**Figure 13.** Time evolution of *Tgas* determined based on the CO rotational band using the Boltzmann method (black squares), direct spectral comparison (red circles), and proposed line-ratio method (green triangles). Reproduced with permission from Ref. [41]. Copyright 2014 OAS Publishing.

#### **4. The characterization of the post-discharge area**

The research related to the post-discharge area characterization of the microwave reactor is presented in this section. A special attention was devoted to the investigation of parameters that may favor the fine-tuning of CO2 decomposition. This brings us to the study of the following effects: (i) Ar admixture, (ii) duty cycle, and (iii) discharge pressure. The motivation behind each effect, along with the results obtained, is presented below.

#### **4.1. Ar admixture effect (TALIF)**

2

The influence of Ar atoms in the chemistry of gas discharges containing oxygen species has been extensively studied over the past years (see, e.g., [47, 48]). In this context, it is well known that the Ar metastables (here denoted by Ar*met*) play a key role as internal energy reservoirs for the production of atomic oxygen species via [48]:

$$\begin{aligned} \text{Ar}^{\text{met}} + \text{O}\_2 &\to \text{O}(^3\text{P}\_2) + \text{O}(^3\text{P}\_2) + Ar \\ &\to \text{O}(^3\text{P}\_2) + \text{O}(^1D) + Ar \\ &\to \text{O}(^3\text{P}\_2) + \text{O}(^1\text{S}) + Ar \end{aligned} \tag{16}$$

Furthermore, due to their long lifetime, the Ar*met* may also contribute to the chemistry of the afterglow and post-discharges. These considerations have triggered the TALIF study on O(3*P*2) formation as a function of the Ar admixture. The typical results related to the O(3*P*2) density measurements in the Ar–O2 and Ar–CO2 mixtures are shown in Figure 14. Indeed, these measurements strongly suggest that Ar atoms enhance the O(3*P*2) production. At low Ar admixtures, the strong molecular quenching induced by O2/CO2 should lead to a decrease in the Ar*met* population and less O(3*P*2) produced via Eq. 16. The last observation is supported by theoretical and experimental results obtained in Ar–O2 [47] and Ar–CO2 [49] MSGDs, which show a decrease of Ar*met* (about three orders of magnitude) in the plasma phase as a consequence of the molecular quenching.

**Figure 14.** Density of the O(3*P*2) as a function of the Ar admixture in O2 (solid line) and CO2 (dashed line) mixtures. The mean power applied was 400 W with a 0.7 kHz pulsing frequency and 33% duty cycle. The total gas flow in both mixtures was 700 sccm. The maximum Ar admixture used was 99%.

It is interesting to note that as the Ar content increases from ∼90% to 100%, there is a clear decrease of O(3*P*2) as the source of oxygen (CO2/O2) is also decreasing. This leads to the formation of the maximum in O(3*P*2) production around 90% of Ar admixture. Furthermore, it should be emphasized that the electron density is expected to decrease as the Ar admixture decreases since the power used for the production of electron-ion pairs is diverted into vibrational and rotational heating of the molecular gas. As a result of lower electron density, one might also expect a decrease in the contribution of the ladder-climbing processes (e.g., stepwise ionization) out of the Ar*met*. These considerations (together with low molecular quenching rate at low molecular admixture) may contribute to an increase of *Armet* population around 90% of Ar admixture. Further investigations are required in order to validate this last assumption, namely, the measurement of Ar*met* densities in the post-discharge region of the reactor.

#### **4.2. Duty cycle effect (TALIF)**

The pulse discharge operation mode offers an additional feature that may increase the flexibility of the global dynamic plasma response in microwave discharges. In many experiments, the power interruption was usually found to be suitable for reducing gas heating in the discharge by injecting a large amount of power during the plasma on-time while keeping the average power at low values [23]. In case of microwave plasmas, this effect was already proven to be advantageous for the production of atomic nitrogen and oxygen in N2 and O2 discharges ([23, 50]).

In this work, the influence of the duty cycle (defined as the ratio between the plasma on-time and its repetition period) on the O(3*P*2) production was investigated by TALIF (see Figure 15). These results clearly show an improvement in the O(3*P*2) production as the duty cycle decreases. The CO2-20% Ar gas mixture with a pulsing frequency of 0.5 kHz was used during these measurements.

**Figure 15.** Left scale: Emission of CO<sup>+</sup> <sup>2</sup> (dashed line) and density of O(3*P*2) (solid line). The TALIF data was normalized to the maximum value obtained in Figure 14 and further multiplied by a factor of 30. Right scale: Gas temperature (crosses) calculated via Eq. 14. The mean power applied was 250 W with a 0.5 kHz pulsing frequency. The CO2-20% gas mixture with total flow rate of 700 sccm was used.

It is important to stress that the average power was kept constant. On the other hand, by decreasing the power per pulse as the duty cycle increases, we found a decrease of *Tgas* (calculated via Eq. 14) that follows the O(3*P*2) evolution. Interestingly, the emission of the ultraviolet doublet system of CO<sup>+</sup> <sup>2</sup> (measured in the plasma phase by OES) shows an increase with the duty cycle. These observations suggest that low duty cycles favor the dissociative recombination of CO2, which is responsible for the release of energy in the form of heating into the plasma [23]. There are other factors that may also play an important role in the power modulation effect, namely, the electron parameters and sources of internal energy (e.g., metastable species). For instance, as shown by the theoretical work of Subramonium et al. [51], it can be expected to find higher *Te* values at the leading edge of the plasma on-time as the duty cycle decreases. This result supports the data obtained, considering dissociation via electron impact of O2 as the major source of O(3*P*2).

#### **4.3. Pressure effect (GC)**

2

The pressure effect on the CO2 decomposition was studied by GC (as described in Section 2.3) in this last part of the work. Figure 16 shows the *χ* (calculated via Eq. 11) and *η* (calculated via Eq. 6) evolution in the range of 20–90 Torr at different pulsing frequencies. The values obtained at low pressures are in good agreement with the theoretical calculations performed by Bogaerts et al. [52] for microwave plasmas sustained at 20 Torr. As the pressure increases, our experimental data clearly shows an improvement of the CO2 decomposition with *η* ∼ 35% and *χ* ∼ 20% at 90 Torr. Figure 16 also suggests that higher *η* values can be achieved with further increase of pressure. Unfortunately, such pressure range was limited by the current pumping system. As suggested by Fridman [9], these pressure-dependent results can be understood based on the increase of electron-neutral collision frequency in the plasma phase which favors the vibrational excitation of the asymmetric vibrational mode of CO2 and leads to higher conversion rates. A possible decrease of *Te* with the increase of discharge pressure might also enhance the vibrational excitation of CO2 since less energy is wasted in electronic excitation. This last observation is supported by *Te* measurements recently obtained in CO2–Ar discharges [49]. However, further increase of pressure (eventually reaching the atmospheric range) may lead to higher VT relaxation rates, faster gas heating, and lower efficiencies. In this case, the plasma would act as a heater, which would bring additional problems such as the reverse reactions forming back CO2.

**Figure 16.** Evolution of the dissociation rate and energy efficiency of CO2 as a function of the discharge pressure at different pulse frequencies. The mean power applied was 2000 W with 50% duty cycle. The flow of CO2 was 15 slm (*Em* ≈ 1.9 eV/molecule).

#### **5. Summary and conclusion**

As new solutions to decrease our dependence on fossil fuels are required, the recycling of CO2 into value-added materials may play a critical role in the near future. In this regard, plasma-assisted decomposition offers a practical and efficient way to convert CO2

into valuable products that can be extremely useful in CCU applications. Therefore, and given the necessity to enhance the CO2 conversion in plasma reactors, there is a constant need to improve the knowledge of CO2-containing discharges through the development of experimental diagnostic techniques and the improvement of theoretical models. This chapter explores the microwave surfaguide discharge as one potential example of plasma technology that can be used for CO2 conversion. In order to fully characterize our system and obtain a large set of valuable data, we used three different diagnostics: (i) optical emission spectroscopy, (ii) two-photon absorption laser-induced fluorescence, and (iii) gas chromatography. The first one is particularly useful to study the plasma region of the reactor, while the other two give an important insight on the density of products in the post-discharge region.

As a result of the optical emission spectroscopy measurements (via actinometry) along the discharge tube, a nonuniform distribution of the CO2 dissociation rate is obtained in the gas propagation direction, which also depends on the specific energy input. This method offers a practical way to construct a CO2 "dissociation map" through the discharge volume with the spatial resolution depending only on the collimating optics used in the optical emission measurements. With a proper population model to describe the excitation processes, such a mapping should give rather accurate results that may be particularly valuable for maximization of the CO2 dissociation efficiency in cold plasma discharges.

A special attention has been also given to the measurement of the gas temperature, which is a fundamental parameter to characterize the dissipation of energy through gas heating. In this regard, a simple gas temperature formula based on the line ratio between two rotational peaks of the CO Angstrom rotational emission band was developed. This method is extremely useful if a quick gas temperature estimation without the spectral synthesis is required. It allows for *Tgas* determination, assuming the presence of CO emission from plasma discharge. The applicability of the proposed line-ratio formula is, however, limited by the temperature of about 3000 K, above which the measurement error is rather high (*>*30%). The validity of the proposed approach was additionally approved by the well-known Boltzmann plot, as well as by direct comparison with simulated spectra.

Concerning the post-discharge characterization, the research presented here was mainly focused on the study of parameters that may favor the fine-tuning of CO2 decomposition. As a result of these measurements, the presence of Ar atoms was confirmed to be beneficial for the formation of the ground state atomic oxygen species. This effect was attributed to the high population density of Ar metastable species in the plasma phase at higher Ar admixtures. The gas mixture CO2:Ar = 1:9 was found to be the most efficient proportion to produce the ground state atomic oxygen. On the other hand, the study of the power modulation effect also showed interesting results, namely, the increase of oxygen ground state species at lower duty cycle values. Finally, what is related to the pressure effect, a linear increase of CO2 decomposition was obtained in the range of 20–90 Torr. The maximum value of energy efficiency obtained was about 35% with a reasonable dissociation rate of 20%. The complete set of results obtained in this work are summarized in Table 2.

Overall, the methods explored/developed in this work offer a valuable set of tools that can be applied in microwave and other plasma sources. In addition, the data presented here may also provide a useful input for modeling of the microwave surfaguide-type discharges operating under CO2 gas mixtures. As a long-term perspective, the authors believe that


**Table 2.** Summary of the various results obtained in this work and the corresponding methods used for the plasma characterization

microwave plasmas are among the most promising candidates to obtain an efficient CO2 decomposition. However, further investigations related to understanding of the basic plasma processes are still required in the atmospheric pressure regime. The use of catalytic materials [15] (e.g., to improve selective production of decomposed products) and efficient discharge configurations (e.g., using a reverse vortex flow [17]) may turn out as the next step toward a practical implementation of microwave plasmas targeted for massive CO2 conversion.

### **Acknowledgments**

This work is supported by the Belgian Government through the "Pole d'Attraction Interuniversitaire" (PAI, P7/34, "Plasma-Surface Interaction", Ψ). The authors do appreciate the valuable comments of Prof. Joost van der Mullen (Universite Libre de Bruxelles, Belgium). N. Britun is a postdoctoral researcher of the Fonds National de la Recherche Scientifique (FNRS), Belgium.

### **Author details**

Tiago Silva1∗, Nikolay Britun1, Thomas Godfroid2 and Rony Snyders1,2

\*Address all correspondence to: tiago.dapontesilva@umons.ac.be

1 Chimie des Interactions Plasma-Surface (ChIPS), Universite de Mons, Mons, Belgium

2 Materia Nova Research Center, Mons, Belgium

2

### **References**


2


[49] Silva T, Britun N, Godfroid T, van der Mullen J, Snyders R. Study of Ar and Ar-CO2 microwave surfaguide discharges by optical spectroscopy. Submitted to *J. Appl. Phys.* 2015.

2

