Plasma Technology and Its Relevance in Waste Air and Waste Gas Treatment

Plasma technology is already used in various applications such as surface treatment, surface coating, reforming of carbon dioxide and methane, removal of volatile organic compounds, odor abatement and disinfection, but treatment processes described in this context do not go beyond laboratory and pilot plant scale. Exemplary applications of both non-thermal plasma and thermal plasma should underline the feasibility of scale-up to industrial application. A non-thermal plasma in modular form was built, which is designed for up to 1000 m3·h−1 and was successfully practically tested in combination of non-thermal plasma (NTP), mineral adsorber and bio-scrubber for abatement of volatile organic components (VOCs), odorous substances and germs. Thermal plasmas are usually arc-heated plasmas, which are operated with different plasma gases such as nitrogen, oxygen, argon or air. In recent years steam plasmas were gradually established, adding liquid water as plasma gas. In the present system the plasma was directly operated with steam generated externally. Further progress of development of this system was described and critically evaluated towards performance data of an already commercially used water film-based system. Degradation rates of CF4 contaminated air of up to 100% where achieved in industrial scale.


Introduction
Plasma is the gaseous fourth state of matter, which is created by the supply of energy in addition to the states of solids, liquids and gases, in which charges are separated from each other in a stable form. This meta-stable state can be generated by thermal energy but is mainly generated by supplying electrical energy in technical plasmas. Natural occurrences of plasmas include lightning, aurora borealis, or the corona of the sun [1].
The first experiments with technical plasmas were conducted as early as 1705 by Francis Hauksbee, who generated arcs of light in evacuated glass spheres [2]. These are still known today as decorative plasma spheres or lamps.
A technical application of plasmas was firstly reported in 1857 by Werner von Siemens, who developed an ozonizer to treat drinking water with ozone [3]. The plasma as such was named in 1928 by Irving Langmuir [4].
Today, plasmas are used in many different areas; as examples in households in the form of gas discharge lamps (energy-saving lamps, neon tubes) and plasma televisions, in industrial for surface treatment (plasma polymerization), analysis technology (plasma ashing), medicine (disinfection, wound treatment) and food industry (disinfection) [5][6][7]. The possible applications of plasmas are manifold and have been used in environmental technology since the 1980s. Non-thermal plasmas, for example, were already used in 1984-1986 for the treatment of military waste. [8,9]. Since then research in plasma sciences focused on geometries of plasmas, test plants and electrodes; search for capable material for construction of plants, electrodes, and barriers; power supply; plasma gases and the selection and optimization of catalytic materials as in-plasma catalysis (IPC) or post-plasma catalysis (PPC) [10][11][12]. However, based on recent publications, research dominantly focuses on bench-scale plasmas of very small system volumes (a few mL) and exhaust air flows (a few mL·min −1 to L·min −1 ) [13,14]. The scale-up factors and operating costs derived from this scale are mostly doubtful or far beyond economically interesting applications. Since plasmas are mostly used for the treatment of hardly degradable pollutants or mixtures, they are in direct competition with incineration processes. Here, the operating costs are essentially defined by the consumption of fossil fuels, which in turn can be directly derived from the temperature difference between clean and crude gas. Modern combustion processes work with a temperature difference of 35 K (in case of crude gas levels above 90 • C) or 50 K (crude gas of ambient temperature). Depending on the quality of the natural gas and neglecting both latent heat losses and energy requirements for compressors and compressed air, this change in temperature corresponds to a specific input energy (SIE) value of 16.0-21.0 kWh·1000 m −3 , which must be used at least as indicator of economic efficiency of non-thermal plasma (NTP) processes.
In this article, the experiences with non-thermal and thermal plasmas on a pilot and industrial scale and their importance for waste air and waste gas purification will be discussed in detail on the basis of own research results of the last years.

Designs and Operational Parameters
As shown above, plasmas can either be generated by supplying thermal energy or electrical energy in the form of kinetic energy. Technical designs of non-thermal plasma include dielectric barrier discharge (DBD), gliding arc, microwave discharge, glow discharge, corona discharge and radio frequency discharge, which were previously described in detail [13,15,16]. The most commonly used method of NTP in pilot-and industrial scale is the dielectric barrier discharge technology. Here, a sufficiently high electrical voltage is applied between two electrodes to ionize the intermediate gas.
To prevent a breakdown by arc discharge, the two electrodes are galvanically separated by a dielectric barrier. After the plasma is ignited, the charges flow in the direction of the electrodes or dielectric barrier, depending on their polarity. Since the dielectric barrier is insulating, the charges cannot flow off and accumulate on the surface. An opposing electric field is thus formed, which causes ceasing of the discharge. By change of polarity, the plasma can be re-ignited. Overall, only surface charges are exchanged in DBD between the electrodes and the dielectric barrier. This corresponds to a displacement current through which electrical power is transferred into the plasma. Changes in polarity take place at frequencies of a few hundred Hz to MHz at up to 30 kV of applied voltage. Only the light electrons in the discharge zone can follow this high frequency, while the heavier cations are hardly accelerated or the momentum changes due to their mass inertia. The electron temperature is thus significantly higher than the heavy particle temperature, which ensures reactivity in the plasma due to the high-energy electrons, and the gas temperature does not increase significantly and remains in the range of the original waste gas temperature. Non-thermal plasmas are therefore not in thermodynamic equilibrium. The effectiveness of the DBE lies in the selective energy coupling into the released electrons, which are accelerated by the electric field to energies between 1 and 10 eV and are subject to a number of reaction mechanisms in the plasma, which were previously described in detail [17,18]. As a consequence of these reaction mechanisms, radicals, negative ions, positive ions, or metastable molecules are formed in humid air as carrier gas out of oxygen, nitrogen and water molecules in different excitation states with different reactivity and lifetime.

Fields of Application
Supported by intensive research in the field of new catalyst development, DBD processes for treatment of surfaces, food, wastewater, wastewater disinfection, and waste gas have been in the focus of research in recent years [7,15,[64][65][66][67]. It is noticeable that especially the application in environmental technology, here especially in the treatment of wastewater, increased. The efficiency of NTP processes for wastewater treatment is mostly simulated by water samples containing dyes, because the analytical evaluation can be done simply by photometric detection. Occasionally, pesticides such as dimethoates [64], pharmaceuticals, phenol or caffeine have also been successfully tested. However, while plasma treatment seems to be a promising route for removal of organic pollutants from water, corresponding NTP degradation pathways are still not understood [64].
Application of DBEs in the field of fiber surface treatment and reforming of methane is also increasing [15]. Applications in waste gas treatment mainly focused on the disinfection of food, food storage rooms, surfaces and VOC abatement [68].

Methane Reforming
Methane reformation is a highly topical issue. Methane is one of the greenhouse gases and is considered as a source for the production of hydrogen. Methane turnover has to been seen in context with plasma-induced reactions of CO 2 , where CO, H 2 , methane and methanol and longer-chain products like ethane, ethene, ethine or propene [69,70] can be formed by dry reforming of methane, CO 2 hydrogenation, CO 2 reduction with water and CO 2 decomposition, and hence, plasma treatment is a promising technology for CO 2 capturing, too [15,67,71,72]. Therefore, many research groups deal with the plasma-assisted production of hydrogen from methane and synthesis gases, considering both non-thermal and thermal plasmas as conversion technology [15,[73][74][75][76].
The use of non-thermal plasma is subject to a limitation: methane is inert and requires at least 1000 • C to cleave the C-H bond [70]. Hence, efficiency in transformation of educts (CO 2 , CH 4 ) and formation of products (H 2 , CO) during dry reforming of methane extremely depends on the SIE values, which are reported in the range of 3333-225,000 kWh·1000 m −3 [15,67,71]. As consequence, economic and ecological advantages cannot be derived directly due to high energy costs and secondary emissions of CO 2 .
In order to use non-thermal plasma for methane reforming at all, catalysts are needed that reduce the starting temperature of methane to 50-300 • C [12]. Montoro-Damas et al. [77] were able to achieve a transformation efficiency of 47.5% in pure methane atmosphere still at high SIE values of 138.900 kWh·1000 m −3 , but considering lower methane concentrations of 2500 ppm, conversion of 32.5% of the methane were achieved at conditions of 50 • C, 0.14 s contact time and a SIE value of 165 kWh·1000 m −3 . Conversely, the inert character of methane can be used to purify methane in biogas or biomass gasification applications by NTP processes [78].

Odor Removal
One application predestined for NTP technology is odor abatement, since low SIE values are required due to the low concentration of odorous components, which means that the process can be used very successfully and cost-effectively. For example, Mizuno et al. (2007) [79] described a reduction in odor of at least 90% in the case of H 2 S emissions from a rubber plant at a contact time of up to 3.3 s and an SIE value of 0.86 kWh·1000 m −3 . In the case of a spice factory, an odor reduction of 80-90% could already be realized at a SIE value of 0.1 kWh·1000 m −3 with an odor concentration in the inlet of 100,000 OU·m −3 . Müller and Zahn (2007) [80] described the treatment of an waste gas from a French fries factory by different DBD process variations. Three of four DBD modifications were able to achieve clean gas levels of less than 500 OU·m −3 at 13,000 OU·m −3 in the crude gas and thus comply with the specifications of the German Technical Instruction on Air Quality Control. The efficiencies were 97.2-98.7%. However, no data on energy consumption was provided. A number of other publications describe emissions of odorous components such as ammonia and dimethylamine as representatives for nitrogenous compounds or hydrogen sulfide, carbon disulfide or dimethyl sulfide as sulfur-containing compounds, resulting in odorous concentrations between 1100 OU·m −3 and 322,000 OU·m −3 . The DBDs in these applications were exclusively laboratory plants with a small volume flow. Hence, enhanced SIE values of 9.2-187.5 kWh·1000 m −3 were used, in one case even 16,700 kWh·1000 m −3 [80][81][82][83][84][85][86][87][88]. Preis et al. (2013) [89] compiled an overview of existing plasma systems in the food industry for the reduction of odorous substances. Odor loads from other industrial applications were comprehensively investigated by Li et al. (2020) [12] at efficiencies of up to 98.7%.

Disinfection and Sterilization
Due to the current pandemic caused by SARS-CoV-2, the demand on disinfecting materials (e.g., protective masks), surfaces and room air has increased. Non-thermal plasmas are well suited for this purpose, as low operational temperature allows application with temperature sensitive materials and surfaces. A variety of so-called reference microorganisms, primarily bacteria, is used for evaluation of the disinfection performance. Microbial disinfection is generated by various mechanisms Classical disinfection processes use a variety of so-called standard organisms, primarily bacteria. Disinfection is based on various mechanisms [16]: (1) The reactive particles of the plasma chemically attack the microorganisms and damage the cell structure in a non-repairable extend [7,[90][91][92][93][94][95]. (2) Electron bombardment creates pores in the cell walls, causing reactive particles to enter the cells directly and damage them [7,94,[96][97][98]. (3) Ions can damage DNA and membrane proteins, impede functions or denature the cell membrane [7,97]. (4) Plasmas form OH radicals and other particles in the presence of water molecules, which can lead to a pH shift within a cell. [99,100]. (5) The reactive particles also shift the redox potential and secondary toxins are formed [7] (6) Electric charge can accumulate on the membrane surface.
Even since the outbreak of the pandemic, some research groups investigated the use of non-thermal plasmas for disinfection. Specifically, Lee et al., 2020 [101], investigated a DBD as a possibility for rapid disinfection of protective masks. Disinfection was primarily achieved by ozone, which is formed by the DBD. Whether plasma disinfection is really effective in deactivating or destroying viruses and in particular SARS-CoV-2 can hardly be proven due to complex cultivation by specific cell lines and handling of highly infectious viruses in comparison to bacteria [102]. Nevertheless, it can be assumed that non-thermal plasmas, which have been proven to have a disinfecting effect on bacteria, can also deactivate viruses.

VOC Abatement
VOCs play one of the most important roles in exhaust air purification, as they are emitted by multiple industries. The treatment of VOCs with DBD is considered and critically described in the reviews of Vandenbroucke et al.  [12]. Although the feasibility of these processes has been demonstrated many times, the treatment of VOCs by NTP proves to be very difficult from an economical point of view, since the concentration of VOCs in the waste gas directly correlates with process' energy requirements. Therefore, the use of NTP processes seems to be either interesting for air pollutants having very low emission limits (e.g., cancerogenic compounds like benzene) or for VOC emission levels just above existing limits. A more advanced purification with higher energy input is generally possible. Various studies evaluated treatment of VOCs and other pollutants at commonly lab-scale conditions with low gas flows and high SIE values [12,16]. An overview of pilot plants with volume flows of 10 m 3 ·h −1 or higher can be found in   [14] as well as in VDI guideline No. 2441 [106].
However, commercial and industrial implementation is very rare and reasons were described in detail [12]. Sum up, the reasons are the lack of adaptability to the state of gas, possible designs, operating costs, formation of byproducts and safety aspects.

Designs and Operational Parameters
As mentioned above, plasma is a gas in which sufficient ionization processes take place by supplying energy, so that the free charges (ions and electrons) determine the properties of the gaseous phase. In contrast to the NTP processes of chapter 2, in thermal plasmas, the applied power is coupled into the system as heat by applying a direct voltage of up to 60 kV with over 95% of efficiency, whereby flame temperatures of up to 12,000 K, in water vapor supported systems of 14,000-22,000 K, can be achieved [107][108][109][110]. The high temperature of the process also allows the transformation of hardly convertible components like CF 4 or SF 6 . The high reaction temperature also allows very short contact times.
The plasma temperature directly correlates with the coupled power and the gas volume flow to be treated and can be easily controlled via these two parameters. Correspondingly, low-volume but high-energy plasma gas flows can be used for treatment of waste gas. In addition to the resulting compact plant design, negative effects such as dilution of present concentrations, cooling effects as well as the formation of undesired byproducts can be efficiently reduced in this way, and the conversion efficiency can be further increased, although the concentration of the secondary products formed depends on a large number of parameters. Due to the high temperature level of the plasma, no carbon deposits occur as in combustion processes for PFCs, which can significantly increase process stability.
However, the efficiency of the process strongly correlates with the coupled power, the geometry of the plasma flame, the addition of additives, the mixing behavior of the waste gas during injection into the plasma zone and energy losses by cooling.
Under technical aspects, relevant plasma gases for the degradation of pollutants are helium, argon, nitrogen, steam, air and oxygen. Most frequently mixtures of argon and nitrogen are used [108,111,112]. In the case of noble gases and nitrogen, the addition of liquid water, water vapor or air is required for oxidative degradation of the pollutants. A detailed consideration of the effects of the various plasma gases on their behavior in the plasma with respect to conversion efficiency and the formation of secondary intermediates is given in a report by Dobslaw et al., 2019 [110]. Here, water vapor plasma proved to be the most efficient plasma form in the conversion of CF 4 (see also Section 6.1). More recently, H 2 and CO 2 have also become increasingly important as plasma gases for the production of synthesis gas [113][114][115].
The advantages of thermal plasmas are a high heat flow, high reaction temperatures, high energy densities, high concentrations of reactive components and thus high transformation efficiency with short contact times. Demand-oriented control, compact design and short start-up times are also possible. The high temperatures in the reaction chamber allow a rapid conversion even of components that are difficult to degrade. By controlling the temperature gradient in the reaction chamber, the coupled power and the addition of potential additives, both the transformation rate and the suppression of undesired byproducts can be accelerated. Furthermore, the demand on supplied gas and liquid is less extensive compared to combustion. The HCl or HF resulting from the treatment of CFCs, CHCs or PFCs can be easily eliminated by a subsequent alkaline scrubber.
On the other hand, there are disadvantages such as the formation of undesired byproducts, especially when oxygen is added as an additive (NOx, COF 2 , CO, etc.), the need for cooling concepts for the reaction chamber to prevent thermal damage to the torch, and a comparatively high electrical power requirement, which strongly depends on the plasma source used as well as the pollutant to be decomposed. Energetically optimized steam plasmas, for example, have a SIE value of approx. 2000-2200 kWh·1000 m −3 for the hardly degradable CF 4 [116][117][118].

Plasma Gases and Their Effects
As listed before, mixtures of argon and nitrogen are used as the most common technical plasma gases. In recent years, steam plasmas in particular have gained increasing relevance. Water can be offered to the plasma torch either as liquid or as externally generated steam. The advantages of water as a plasma gas compared to noble gases are, apart from clear economic aspects [119], the physical advantages. For example, the OH radicals produced have a high oxidative effect, and water vapor can provide higher energy densities due to its comparatively high c p value than other plasma gases (see Figure 1).

Plasma Gases and Their Effects
As listed before, mixtures of argon and nitrogen are used as the most common technical plasma gases. In recent years, steam plasmas in particular have gained increasing relevance. Water can be offered to the plasma torch either as liquid or as externally generated steam. The advantages of water as a plasma gas compared to noble gases are, apart from clear economic aspects [119], the physical advantages. For example, the OH radicals produced have a high oxidative effect, and water vapor can provide higher energy densities due to its comparatively high cp value than other plasma gases (see Figure 1). Oxidative processes require the presence of oxygen molecules or atoms. An alternative to steam would therefore be oxygen as plasma gas, but pure oxygen leads to corrosive effects on electrodes and other components of the torch and reactor. As a consequence, the lifetime of such reactors is considerably reduced. Therefore, the use of water-supported plasmas is preferable. If steam is generated and fed in externally, it is advantageous to avoid cooling of the plasma flame by the evaporation enthalpy to be applied, thus avoiding negative effects on cleaning efficiency due to reduced temperatures.
Especially in the field of pyrolysis, Nishikawa et al., (2004) [121] have proven in experiments with charcoal that steam as plasma gas promotes the formation of a higher quality synthesis gas than argon or oxygen.
In the following, the different ways of introducing water are summarized. This is based on the usual system design, consisting of a plasma torch, a gas mixing chamber, a reaction chamber and a quench for gas cooling: (1) Injection into the plasma chamber upflow the quench [116,122].
(3) Injection after quenching in the direction of the inner reactor wall [116,122]. (4) Transmission of the cold plasma gas through an upstream water tank [108,123]. Oxidative processes require the presence of oxygen molecules or atoms. An alternative to steam would therefore be oxygen as plasma gas, but pure oxygen leads to corrosive effects on electrodes and other components of the torch and reactor. As a consequence, the lifetime of such reactors is considerably reduced. Therefore, the use of water-supported plasmas is preferable. If steam is generated and fed in externally, it is advantageous to avoid cooling of the plasma flame by the evaporation enthalpy to be applied, thus avoiding negative effects on cleaning efficiency due to reduced temperatures.
Especially in the field of pyrolysis, Nishikawa et al. (2004) [121] have proven in experiments with charcoal that steam as plasma gas promotes the formation of a higher quality synthesis gas than argon or oxygen.
In the following, the different ways of introducing water are summarized. This is based on the usual system design, consisting of a plasma torch, a gas mixing chamber, a reaction chamber and a quench for gas cooling: (1) Injection into the plasma chamber upflow the quench [116,122].
(3) Injection after quenching in the direction of the inner reactor wall [116,122]. (4) Transmission of the cold plasma gas through an upstream water tank [108,123]. (5) Passing the hot plasma gas through a downstream water tank [108,123].
The respective advantages and disadvantages are explained by Dobslaw et al. (2020) [16]. Common to all mentioned methods is that they work with liquid water. Commercially, variant 3 (injection in the direction of the inner reactor wall) is most frequently used, since it allows a long lifetime with low risk of deposits and, with regard to CF 4 , a high degree of efficiency.
However, it is energetically more efficient to use water as steam. There are different process technologies to deal with this issue. Watanabe et al. [123][124][125] have developed a plasma source in which liquid water is vaporized in the anode area. Advantages of this technique are mainly the high-volume flow of up to 0.5 L·h −1 and the cooling of both the electrodes and the reaction chamber. However, it turned out to be a disadvantage that the liquid water flow led to a plasma cooling (T max < 3000 K), and the evaporation of the water caused cavitation effects and damage of the torch. These negative effects led to a considerably reduced lifetime of less than 100 h.
Hrabovski et al. (2011) [120] tried to avoid the problem of corrosion by constructive adaptation. On the one hand, argon was introduced into the plasma chamber as a protective and plasma gas, and on the other hand, the inner walls were made of semi-permeable sintered metal and permeated by water. The steam produced by the heat was then introduced into the reaction chamber and provided sufficient cooling. In addition, the anode was designed as a rotating copper disc, which means that the discharge runs permanently along the edge, thus reducing local corrosion. Although the lifetime could be increased in this way, it requires an additional motor to drive the copper disc.
In a next step, Watanabe [126] used a hafnium electrode in a copper tube as cathode. A copper cylinder served as anode. The externally supplied water is vaporized by the waste heat at the anodes and fed into the plasma through the anode nozzle before the plasma gas escapes. With this process, an expensive shielding gas and an external vaporizer can be dispensed with. Disadvantages of this design are that no defined water vapor flow can be set, only a defined exhaust air flow is possible, strong corrosion effects occur and a high energy input is required. Glocker et al. (2000) [127], on the other hand, works with external steam generation. The volume flow of the water vapor can thus be precisely adjusted and is passed by the cathode that is surrounded by argon, thus entering the plasma flame. The advantages of this process are the long lifetime of the cathode and the controllability of the steam flow. Disadvantages are the comparatively moderate lifetime of the anode, the expensive equipment for controlling the different gas flows and the necessary external cooling of the electrodes. This system configuration served as the basis for the further development described in later chapters.

Environmental Application
Thermal plasmas have been used in engineering applications since the 1980s. In the 1990s, they were also used for the first time in waste gas treatment. Since then, thermal plasmas have been successfully used for the treatment of gaseous emissions of VOCs such as acetone [128], decanol [129] and PCBs as well as PFCs (CF 4 , C 2 F 6 , C 3 F 8 , C 4 F 8 , NF 3 , SF 6 , CHF 3 , HFC-23, HFC-134a, HFC-32) [107,116,123,[130][131][132][133][134]. Recently, thermal plasmas have also been used for the treatment of aqueous wastes [129,135] and for solids detoxification [126,136]. Due to the high energy demand described above, their use is restricted from an economic point of view in particular to exhaust air and waste streams that are very difficult to treat, such as the treatment of PFCs and CFCs.
Perfluorinated hydrocarbons such as CF 4 , C 2 F 6 , C 3 F 8 , C 4 F 8 as well as CHF 3 , NF 3 and SF 6 occur as greenhouse relevant emissions in the semiconductor industry, in aluminum extraction as well as in the waste management industry and have also been used in the past in high-voltage equipment, in magnesium die-casting as well as in the filling of car tires. More recently, PFC emissions from the electrolysis of rare earth fluoride/oxide systems such as neodymium and dysprosium have become increasingly important [137]. Total global emissions of CF 4 , C 2 F 6 and SF 6 in the two main sectors of the semiconductor industry and aluminum production were 42 Mt CO 2,eq ·a −1 and 25 Mt CO 2,eq ·a −1 (as of 2012) for CF 4 and C 2 F 6 , respectively, and 35 Mt CO 2,eq ·a −1 of additional SF 6 in the semiconductor industry [137,138]. Further applications are surface cleaning and surface degreasing in the fields of electronics, electrical engineering and aerospace, high-voltage systems in heavy industry, filling gases in soundproof windows and car tires, magnesium die casting, various chemical syntheses in the chemical industry, as oxidizing agent for special fuels in the aerospace industry and the production of oscillating F 2 high-power lasers [139].

NTP Plasma
Tests with NTP processes were carried out with an in-house developed DBD systems. Starting from a commercial product (Figure 2a), the short lifetime of reactive plasma components and the positive effect of catalytic adsorbers on cleaning efficiency were considered by developing a new housing with integrated PPC in the form of a drawer system (Figure 2b,c). As catalytic adsorbers, five different nanoscale clay minerals were tested according to Table 2. In pre-tests, these minerals were qualified with respect to their adsorption and desorption properties against selected waste gas components (methane 1 g·m −3 ; n-butanol 0.1 g·m −3 ; ammonia 0.1 g·m −3 ). The test setup is shown schematically in Figure 3. A mixture of ammonia, n-butanol and methane at volume flows of 300, 500 and 700 m 3 ·h −1 was used as artificial waste gas in the NTP process. The geometrical parameters were previously described [31]. In parallel, a second system for spectroscopic analysis of the plasma discharge was developed by installing a polycarbonate window in the stack and in the front of the housing. Table 2. Overview of the clay minerals used.

cristobalite and muscovite in total
The flow profiles in the drawer-equipped NTP-PPC combination were simulated with the software Solidworks 2014, and the results of this simulation were incorporated into the design of a pilot plant as shown in Figures 4 and 5. This pilot plant, also an NTP-PPC combination, was built in a container design and equipped with a downstream installed drum filter with mineral adsorber GA2, an activated carbon coated honeycomb adsorber for ozone elimination and a bioscrubber with associated periphery built outside the container (bioscrubber only in phase I, see Figure 6). A cyclone and a commercial filter for the elimination of aerosols and dust were installed upstream. The selection of GA2 was based not only on the results of the pre-tests but also on the aspects of the commercial exploration grade of the mining site and the political stability at the exploration site. With the installed ventilation capacity in the pilot plant, waste gas flows of up to 1000 m 3 ·h −1 could be treated. Both laboratory and pilot plant could be operated in inline and bypass mode. In the interest of high process stability, the NTP was operated in bypass mode to exclude operational malfunctions due to high humidity or dust loads. The geometrical characteristics of this plant were previously described [14].
Experimental tests of this plant were carried out at three locations, whereby at location I, additional operating phases were simulated by adding various VOCs to the present waste gas. The central operating parameters ar4e summarized in Table 3.
Sampling points for gas analysis were available in the crude gas, after NTP stage, after mineral adsorber, after honeycomb and in the clean gas after the optional bioscrubber.

Thermal Plasma
The evaluation of the thermal plasma was carried out at using an industrial-scale test system consisting of a gas supply, external vaporizer, power source, plasma torch, gas mixing chamber, reaction chamber, quench and an alkaline scrubber installed downstream. Figure 7 shows the installation in operation with an installed radiation-cooled reaction chamber (right) and supplemented gas analysis (left). The turnover of CF 4 as model pollutant in concentrations between 2790 ppm and up to 80,000 ppm at waste gas flows of 50-300 slm, at electrical power input of 12-20 kW, was tested using different plasma gases. As plasma gases N 2 , O 2 , air and water vapor were used. In addition, three reaction chamber geometries (waterfall film, radiation-cooled reaction chamber, recuperative reaction chamber) and two nozzle geometries (axially expanding nozzle, radially expanding Laval nozzle) were tested. installation in operation with an installed radiation-cooled reaction chamber (right) and supplemented gas analysis (left). The turnover of CF4 as model pollutant in concentrations between 2790 ppm and up to 80,000 ppm at waste gas flows of 50-300 slm, at electrical power input of 12-20 kW, was tested using different plasma gases. As plasma gases N2, O2, air and water vapor were used. In addition, three reaction chamber geometries (waterfall film, radiation-cooled reaction chamber, recuperative reaction chamber) and two nozzle geometries (axially expanding nozzle, radially expanding Laval nozzle) were tested.

Combined NTP/Bio Process for the Treatment of Complex Waste Gases
As described before, a combined system of NTP and subsequent biological treatment for energyefficient treatment of complex waste gases such as those occurring in animal husbandry, animal meal drying, sewage sludge drying, waste composting, fermentation residue treatment and other industrial fields (50-100 organic components in total concentrations of some 100 mg C•m −3 , methane contents of 10-150 mg C•m −3 and odor concentrations of sometimes more than 20,000 OU•m −3 ) should be achieved. The process of development can be subdivided into eight steps as follows: 1. Optical characterization of the plasma discharge as an indication of the formation of reactive components and the expected cleaning efficiency; 2. Development of a suitable catalytically active mineral adsorber; 3. Design and construction of a first prototype; 4. Testing of the prototype; 5. Flow simulation of the first prototype;

Combined NTP/Bio Process for the Treatment of Complex Waste Gases
As described before, a combined system of NTP and subsequent biological treatment for energy-efficient treatment of complex waste gases such as those occurring in animal husbandry, animal meal drying, sewage sludge drying, waste composting, fermentation residue treatment and other industrial fields (50-100 organic components in total concentrations of some 100 mg C·m −3 , methane contents of 10-150 mg C·m −3 and odor concentrations of sometimes more than 20,000 OU·m −3 ) should be achieved. The process of development can be subdivided into eight steps as follows: 1.
Optical characterization of the plasma discharge as an indication of the formation of reactive components and the expected cleaning efficiency; 2.
Development of a suitable catalytically active mineral adsorber; 3.
Design and construction of a first prototype; 4.
Testing of the prototype; 5.
Flow simulation of the first prototype; 6.
Development of a pilot plant for waste gas flows up to 1000 m 3 ·h −1 , 7.
Testing of the pilot plant; 8.
Evaluation of experimental results and process stability with derivation of further development potentials.

Optical Characterization
The optical analysis of the discharge homogeneity of the plasma is an important indicator for the expected cleaning efficiency of the process, since only a homogeneous plasma discharge allows an optimized contact of the waste gas pollutants passing through with the generated reactive plasma components. The homogeneity of the plasma is influenced by the applied frequency of the AC voltage as well as by the applied current. The homogeneity of the plasma was therefore measured by varying these two parameters in the test system with spectroscopic access. Figure 8 shows the DBD-stack (left) converted for spectroscopic access with the occurring discharge (right). The frequency of the discharge was varied in the range of 300-750 Hz, the current in the range of 7-19 A. Stable operating conditions with associated discharge characteristics are shown in Table 4. It can be seen that, as expected, the homogeneity and intensity of the plasma discharge increases with increasing current. Considering aspects of energy optimization of the system, it is interesting to note that a higher discharge homogeneity can be achieved at low frequencies. For further experiments, the operating condition at 11 A and 333 Hz was identified as the best set of parameters, since this is where the highest plasma homogeneity occurred. Additional experiments with variable gas flow at 11 A and 333 Hz did not show any effect on the homogeneity and stability of the plasma (data not shown).
Macro photos of the plates for a more detailed view of the discharges resulted in the situation shown in Figure 9. In the left figure, discharge channels are clearly visible showing a pronounced discharge foot at the ceramic barrier and a slender head at the electrode. Due to the use of a rough or structured ceramic material, the micro discharges occur preferably stationary at material elevations (right). These material elevations caused by the surface roughness represent fixed release points for the discharge, which is reflected in a higher light intensity at the discharge point. Table 4. View on plasma homogeneity of some selected combinations of current and frequency [140].
6. Development of a pilot plant for waste gas flows up to 1000 m³•h −1 , 7. Testing of the pilot plant; 8. Evaluation of experimental results and process stability with derivation of further development potentials.

Optical Characterization
The optical analysis of the discharge homogeneity of the plasma is an important indicator for the expected cleaning efficiency of the process, since only a homogeneous plasma discharge allows an optimized contact of the waste gas pollutants passing through with the generated reactive plasma components. The homogeneity of the plasma is influenced by the applied frequency of the AC voltage as well as by the applied current. The homogeneity of the plasma was therefore measured by varying these two parameters in the test system with spectroscopic access. Figure 8 shows the DBD-stack (left) converted for spectroscopic access with the occurring discharge (right). The frequency of the discharge was varied in the range of 300-750 Hz, the current in the range of 7-19 A. Stable operating conditions with associated discharge characteristics are shown in Table 4. It can be seen that, as expected, the homogeneity and intensity of the plasma discharge increases with increasing current. Considering aspects of energy optimization of the system, it is interesting to note that a higher discharge homogeneity can be achieved at low frequencies. For further experiments, the operating condition at 11 A and 333 Hz was identified as the best set of parameters, since this is where the highest plasma homogeneity occurred. Additional experiments with variable gas flow at 11 A and 333 Hz did not show any effect on the homogeneity and stability of the plasma (data not shown).
Macro photos of the plates for a more detailed view of the discharges resulted in the situation shown in Figure 9. In the left figure, discharge channels are clearly visible showing a pronounced discharge foot at the ceramic barrier and a slender head at the electrode. Due to the use of a rough or structured ceramic material, the micro discharges occur preferably stationary at material elevations (right). These material elevations caused by the surface roughness represent fixed release points for the discharge, which is reflected in a higher light intensity at the discharge point.     Further spectroscopic analyses of the total discharges of the stack and the individual discharges as well were performed, showing no differences in the present discharge spectrum. Having a detailed look at the present electronic transitions, it was found that for DBD in air only the transition of nitrogen in the 2nd positive system could be observed (see Figure 10). In this excited level, nitrogen molecules have only a short lifetime and return to the basic level by light emissions. All other levels are excited to the ground state by impacts and thus non-radiative decay. Furthermore, a linearly increasing spectral absorption of ozone could be observed with increasing coupled power from 500 W on. Based on Figure 11, an ozone yield of 1 g ozone per 18.7 Wh of power input can be determined and, therefore, is in the upper range of the literature (1 g ozone per 20 Wh). Further spectroscopic analyses of the total discharges of the stack and the individual discharges as well were performed, showing no differences in the present discharge spectrum. Having a detailed look at the present electronic transitions, it was found that for DBD in air only the transition of nitrogen in the 2nd positive system could be observed (see Figure 10). In this excited level, nitrogen molecules have only a short lifetime and return to the basic level by light emissions. All other levels are excited to the ground state by impacts and thus non-radiative decay. Furthermore, a linearly increasing spectral absorption of ozone could be observed with increasing coupled power from 500 W on. Based on Figure 11, an ozone yield of 1 g ozone per 18.7 Wh of power input can be determined and, therefore, is in the upper range of the literature (1 g ozone per 20 Wh).

Catalytically Active Mineral Adsorber
Multiple references in literature showed a significant increase in purification efficiency of the NTP in case of both adsorptive and catalytic effects of the installed PPC. In order to eliminate the potential risk of fire in the large-scale adsorber by glow discharge or induced flying sparks, nanoscale mineral materials were tested with regard to their adsorption capacity against methane, n-butanol and ammonia in realistic concentrations in continuous flow, which indirectly includes adsorptionrelevant transport limitations. The results of these adsorption tests are shown in Table 5. Table 6 shows a ranking of the adsorbers used with regard to their adsorption capacity compared to the artificial waste gas components. Based on this ranking, the further use of GA 4 should be favored. However, since the degree of commercial exploration as well as the political stability at the exploration sites strongly regulated the availability and supply safety, the further tests were finally carried out with the material GA 2. The GA 1 material, which is also commercially readily available, was not pursued further, as exposure of this material to plasma and ammonia generated a white precipitate, which turned out to be ammonium sulfate with traces of pyrite and calcite and clogged the package of the adsorber by forming a top layer. Furthermore, formation of critical secondary intermediates like N2O for GA 2 was lower than for GA 1. Table 5. Adsorption characteristics of mineral adsorbers (mg•kg −1 ) at fixed crude gas levels (g•m −3 ).

Catalytically Active Mineral Adsorber
Multiple references in literature showed a significant increase in purification efficiency of the NTP in case of both adsorptive and catalytic effects of the installed PPC. In order to eliminate the potential risk of fire in the large-scale adsorber by glow discharge or induced flying sparks, nanoscale mineral materials were tested with regard to their adsorption capacity against methane, n-butanol and ammonia in realistic concentrations in continuous flow, which indirectly includes adsorption-relevant transport limitations. The results of these adsorption tests are shown in Table 5. Table 6 shows a ranking of the adsorbers used with regard to their adsorption capacity compared to the artificial waste gas components. Based on this ranking, the further use of GA 4 should be favored. However, since the degree of commercial exploration as well as the political stability at the exploration sites strongly regulated the availability and supply safety, the further tests were finally carried out with the material GA 2. The GA 1 material, which is also commercially readily available, was not pursued further, as exposure of this material to plasma and ammonia generated a white precipitate, which turned out to be ammonium sulfate with traces of pyrite and calcite and clogged the package of the adsorber by forming a top layer. Furthermore, formation of critical secondary intermediates like N 2 O for GA 2 was lower than for GA 1.

Design, Construction and Efficiency of NTP-PPC System
GA 2 as selected material was further tested as PPC in combination with an upstream NTP in treatment of methane, n-butanol and ammonia as references. Performance was quantified by FID analysis (type 2010T, Testa GmbH, München, Germany) in the case of methane and n-butanol and by FTIR analysis for all three components (Antaris IGS, Thermo Fisher, Dreieich, Germany). The latter method also allowed the quantification of reaction products and secondary emissions.
The concentrations of the original pollutant and the reaction products are shown in Figure 12 (methane), Figure 13 (n-butanol) and Figure 14 (ammonia) for a fixed gas flow rate of 300 m 3 ·h −1 (CH 4 , NH 3 ) and 500 m 3 ·h −1 (n-butanol), respectively.   As shown in Figure 12, no significant turnover of methane was observed under the current process conditions. Only in case of a power input of 1.5 kW and a frequency of 400 Hz higher levels of methanol occurred, indicating starting of methane conversion. However, the negligibly low efficiency at a SIE value of approx. 6.7 kWh·1000 m −3 indicates both an extremely questionable and non-energy-efficient conversion of methane, especially since the power consumption continues to rise with increasing methane concentrations and has to compete with incineration processes (SIE values approx. 16-21 kWh·1000 m −3 ).
Energetically, the conversion of ammonia or n-butanol proves to be more favorable (see Figures 13 and 14). Efficiencies of 25-45% and 30-62% were achieved at concentrations of 49-69 ppm NH 3 and 50 ppm n-butanol and SIE values of 3.0-8.0 kWh·1000 m −3 and 3.3-6.7 kWh·1000 m −3 , respectively. However, performance remained well below expectations.   Bars are mean values of 72 individual measurements with a relative measurement deviation within the series of less than 2.5%, equivalent to less than 1.5% in total.

Flow Simulation of the NTP-PPC Process
According to the low performance levels, the flow profiles in the plasma stage as well as in the adsorber drawer were simulated for both inline and bypass operation at gas flows of 300, 500, and 700 m³•h −1 in inline operation and for the same gas flows with an additional auxiliary air flow of 65 m 3 •h −1 in bypass mode. Figure 15 shows the situation in bypass operation mode with 300 m³•h −1 of waste air and 65 m³•h −1 of auxiliary air. Here, the waste air is supplied from the left, while the auxiliary air is supplied from the top. The simulation profiles show on the one hand that the waste air does not flow through the bed but flows quantitatively around the adsorber package via the small gaps on Figure 14. Comparison of the cleaning efficiency of the mineral adsorbers GA 1 and GA 2 versus performance without PPC and MnO 2 as reference catalyst in case of n-butanol at a gas flow of 500 m 3 ·h −1 in bypass operation mode. Additionally, the corresponding SIE and EC values are presented. Bars are mean values of 72 individual measurements with a relative measurement deviation within the series of less than 2.5%, equivalent to less than 1.5% in total.

Flow Simulation of the NTP-PPC Process
According to the low performance levels, the flow profiles in the plasma stage as well as in the adsorber drawer were simulated for both inline and bypass operation at gas flows of 300, 500, and 700 m 3 ·h −1 in inline operation and for the same gas flows with an additional auxiliary air flow of 65 m 3 ·h −1 in bypass mode. Figure 15 shows the situation in bypass operation mode with 300 m 3 ·h −1 of waste air and 65 m 3 ·h −1 of auxiliary air. Here, the waste air is supplied from the left, while the auxiliary air is supplied from the top. The simulation profiles show on the one hand that the waste air does not flow through the bed but flows quantitatively around the adsorber package via the small gaps on both sides between the respective guide rail and the drawer, so that the adsorber performs neither as catalyst nor as adsorber. On the other hand, it can be seen that even with this small waste air flow, the pressure loss distribution in the system leads to a backflow of the waste gas into the stack, where it causes strong turbulence above the stack. The backflow of the waste gas leads to a second contact with the plasma stage, which probably leads to the higher efficiency compared to the sole plasma operation mode.
In addition, it was found that the amount of mineral adsorber as used in the drawer model is unsuitable for industrial use, as the drawer system would be unmanageable with waste gas flows of 1000 m 3 ·h −1 or higher. The advantage of the very short gas flow distance between plasma module and adsorber therefore had to be dropped as well.

Design, Construction and Efficiency of the Pilot-Scale NTP-PPC-Bioscrubber System
In view of the disadvantages mentioned above, a further concept was developed, consisting of NTP and a drum containing the mineral adsorber in downstream. Through separation of both stages, possible backflow effects were excluded, and mixing of the waste gas with auxiliary gas during bypass operation was improved in the NTP stage built as case-in-case model. Side flows were further reduced by reducing distances and installing seals. This improved housing also allows easy switching between inline and bypass operation. Flow simulations of the NTP housing underlined the improved gas mixing and were previously described in detail [31]. The plant was operated during three operational phases according to three different waste gas sources according to Table 3 and extended by a bioscrubber as final cleaning stage during the first operating period in order to eliminate the partially oxidized components in an energy-efficient biodegradative way [31]. The process time of this first phase is subdivided into six stages, which are characterized by an artificial addition of VOCs to the original emissions from the sludge centrifugation.
In stage 1, crude gas concentrations of 30-50 mg C•m −3 were measured. Methane levels ranged between 30-110 ppm, and ammonia was about 0.2 ppm on average. According to the poor results of methane conversion during the pre-tests with the drawer model, little or no methane degradation could be detected. However, removal efficiencies of 100% and 78% were achieved in the degradation of ammonia and odor, respectively.

Design, Construction and Efficiency of the Pilot-Scale NTP-PPC-Bioscrubber System
In view of the disadvantages mentioned above, a further concept was developed, consisting of NTP and a drum containing the mineral adsorber in downstream. Through separation of both stages, possible backflow effects were excluded, and mixing of the waste gas with auxiliary gas during bypass operation was improved in the NTP stage built as case-in-case model. Side flows were further reduced by reducing distances and installing seals. This improved housing also allows easy switching between inline and bypass operation. Flow simulations of the NTP housing underlined the improved gas mixing and were previously described in detail [31]. The plant was operated during three operational phases according to three different waste gas sources according to previous Table 3 and extended by a bioscrubber as final cleaning stage during the first operating period in order to eliminate the partially oxidized components in an energy-efficient biodegradative way [31]. The process time of this first phase is subdivided into six stages, which are characterized by an artificial addition of VOCs to the original emissions from the sludge centrifugation.
In stage 1, crude gas concentrations of 30-50 mg C·m −3 were measured. Methane levels ranged between 30-110 ppm, and ammonia was about 0.2 ppm on average. According to the poor results of methane conversion during the pre-tests with the drawer model, little or no methane degradation could be detected. However, removal efficiencies of 100% and 78% were achieved in the degradation of ammonia and odor, respectively.
After 12 days, the crude gas was spiked with 9-15 mg C·m −3 styrene (stage 2), because concentration levels of VOCs in the crude gas from the sludge centrifugation were too low. The styrene was almost completely removed from the waste gas. Benzaldehyde, heptanal and hexanal were detected as more hydrophilic conversion products.
An addition of super gasoline increased the total carbon levels to up to 300 mg C·m −3 in maximum (stage 3). In contrast, VOC levels in the clean gas after DBE were less than 40 mg C·m −3 .
In a further step, the system was charged with a mixture of 90% styrene and 10% ethanol (stage 4), thus achieving an average concentration of 50-70 mg C·m −3 and a maximum level of 350 mg C·m −3 in the crude gas. In the clean gas levels of 0-20 mg C·m −3 were measured during this test series.
Since the laboratory tests showed quite good results in the degradation of n-butanol, n-butanol was also added to the crude gas in step 5. With an addition of 40-60 mg C·m −3 n-butanol, degradation efficiencies of up to 60% and, therefore, about 30% higher than during laboratory pre-tests (efficiency was 30% in maximum) were achieved. Measurements after the biological stage even showed degradation rates of 93-100%.
Finally, 2-butoxyethanol was added in step 6. Due to its low vapor pressure, only low C loads could be evaporated, which meant that only increases in gas concentrations of 15-20 mg C·m −3 were achieved. Accordingly, an average of 10 mg C·m −3 of the carbon concentration could be eliminated by plasma treatment.
Results of the operational periods II and III were previously described in detail [14]. Nevertheless, some additional aspects of operation during period III (treatment of dryer waste gas) should be pointed out. The combined system without bioscrubber was installed for several months at the solid waste fermentation plant in Leonberg to treat waste gas emissions of the belt dryer for drying of the fermentation residues of the fermentation process of solid wastes in a bypass process. The performance of the system was evaluated by continuously operating FID and FTIR analyses and discontinuously operating olfactometric and GC-MS analyses. However, latter tests were stopped after a short time due to the low loads of emissions. Unfortunately, long-term continuous analysis of the process was not possible, because the plant had to struggle with different technical difficulties. To solve these problems, the control system had to be partially deactivated, which, however, prevented the plasma from being ignited automatically. Due to the wet and cold weather in the winter months, the relatively high humidity of the fresh air supply led to recurring system malfunctions. In addition, at the end of the test period, there was a longer shutdown of the plant, as a fire broke out inside the fermentation plant. An exemplary section of the analytical data is presented in Figure 16.
The VOC removal efficiency during the time period of Figure 16 ranged within 10-50%, in individual cases, 100% was also achieved. While the removal of VOCs as total carbon and as various individual components could not provide satisfactory results, which is partly due to the discontinuous test procedure, the olfactometric investigations showed a reproducible positive effect of the plasma-supported treatment of the waste gas. Permanent removal of odor by more than 95% was achieved.
In a subsequent phase IV, the sole NTP stage was connected downstream of a biotrickling filter for the treatment of styrene emissions in order to efficiently sterilize the clean gas. This study was previously described by Helbich et al. (2020) [141]. In summary, a disinfection of more than 3 log units was achieved. This finding also underlines the possibility of using NTP processes for disinfection. was achieved.
In a subsequent phase IV, the sole NTP stage was connected downstream of a biotrickling filter for the treatment of styrene emissions in order to efficiently sterilize the clean gas. This study was previously described by Helbich et al., (2020) [141]. In summary, a disinfection of more than 3 log units was achieved. This finding also underlines the possibility of using NTP processes for disinfection. Figure 16. Course of VOC levels in crude and clean gas (after activated carbon filter) in the belt dryer waste gas of the Leonberg solid waste fermentation plant. NTP is turned on during yellow shaded time periods.

Experimental Evaluation, Costs and Development Potentials
The difficulties that have already been described in practical operation show that there is still room for improvement in various areas. The necessity of bypass operation of the NTP stage in case of humid or dust-laden waste gas has a particular effect on the technical performance. While an efficient dust removal can be achieved by pre-treatment stages, a reduction of the air humidity is only possible either by drying, intermediate heating or by adding dry air. In view of the high waste gas flows, all approaches are associated with high costs, which is why the loss of short-living reactive components out of the plasma discharge must be accepted in order to achieve a cost-efficient reduction of the waste gas humidity by a cost-effective addition of fresh air. Fresh air is led directly through the plasma and serves as a carrier gas for the reactive products to mix them into the waste gas of large volume flows. Besides the humidity itself, a rising water content of the barrier sheets at high humidity process conditions has a negative effect on the process stability, since the dielectric constant of the sheet can decrease by several log levels, leading to increased flashover formation. Additionally, a residual structuring of the plates shows a significant influence on the homogeneity of the plasma discharge. Current in-house research therefore focuses on even, moisture-repellent barriers with preferably catalytic and adsorptive properties and lower weight than the current plates in order to achieve not only a simplified handling but also lower manufacturing costs. Especially the barrier plates cause a high proportion of the constructive costs. If the approach of the process combination of NTP-PPC-bioscrubber is successfully pursued, i.e., the NTP stage is used for partial oxidation of the mostly lipophilic waste gas components and the same is mineralized energy-efficiently by a compact biological process in downstream, the process combination already represents an interesting process alternative even with the current constructive costs for the barriers. In a study by Dobslaw et al. (2017) [31] a comparative calculation with an sole NTP-PPC process has already been carried out based on calculations of Martin et al. (2008) [142]. Extrapolated to a waste gas flow of 20,000 m 3 ·h −1 , the cost advantage of the combination with biological aftertreatment could be estimated at EUR 71,100 in constructive costs and in the range of EUR 37,250 in operating costs per year, equivalent to a cost reduction of specific constructive costs of 3.56 €·h·m −3 and specific operating costs of EUR 0.24-0.94·1000 m −3 . If the development of new barriers succeeds, the constructive costs can be further reduced by a medium five-figure amount, and the operating costs can be reduced by approx. 20% through the inline use of the NTP stage.

Ecological Aspects and Sustainability
In addition to the economic saving potential of NTP processes compared to conventional combustion processes, NTP processes also have a clear advantage from a sustainability perspective. In the case of the NTP processes, CO 2 emissions according to Table 7 result quantitatively from the provision of the electric current for the generation of the plasma discharge (67.0%), while in the case of combustion CO 2 , emissions are produced both by the combustion of natural gas as fuel (42.8%) and by the formation of nitrous oxide as secondary pollutant (41.6%). A direct quantitative comparison of the two processes shows that the NTP process has a savings potential of almost 40 t CO 2 ,eq·a −1 per 1000 m 3 ·h −1 of gas flow or 67.3% compared to the conventional combustion process.

Optimization of Thermal Plasma Processes
According to the state of the art as described above, thermal plasmas have four major problems: 1. Insufficient efficiency due to heat losses; 2.
Insufficient efficiency at high waste gas flows and thus limited SIE values; 3.
Quality and quantity of byproducts; 4.
High losses of coupled power by high heat losses.
In order to meet these problems, test series with different reactor geometries and plasma gases were carried out based on the water vapor plasma developed by Glocker et al. [127] with the objective that the reaction temperature in the plasma zone should be kept above 2500 K for large plasma volumes.

Effect of Geometry and Plasma Gases
Within these tests, two fundamentally different reaction chamber geometries (waterfall film, radiation-cooled chamber), two nozzle geometries (axially expanding nozzle or radially expanding Laval nozzle) as well as four different plasma gases (H 2 O, N2, O 2 , air) were initially investigated, and the water vapor plasma was selected as the most promising plasma gas and the radiation-cooled reactor geometry as the preferred geometry. Test series on the influence of the volume flows of waste gas, plasma gas, coupled electrical power, concentration of CF 4 and nozzle geometries showed a strongly differentiating result.
In the case of the plasma gases, O 2 and air, the formation of high concentrations of NOx was shown, which are responsible for the formation of ground-level ozone and acid rain due to their toxicity and relevance for acid rain. Their formation is also of high relevance in the technical system, since due to their acidic character, the consumption of caustic solution in the downstream scrubber rose sharply, thus worsening the economic balance. A corresponding configuration was no longer considered.
Furthermore, the falling film system showed a significantly higher heat loss to the water phase compared to the radiation-cooled approach, which was indirectly shown by a lower cleaning performance compared to CF 4 under comparable conditions. This effect is illustrated in Table 8, which provides a summary of the efficiencies as a function of the injected CF 4 concentrations for the existing configuration of the steam plasma, the new development with a perforated inner shell and the falling film systems based on H 2 O or N 2 plasma. The direct comparison between nitrogen and water vapor plasma in the falling film shows a significantly higher efficiency of the water vapor plasma based on the differences in efficiency. The comparison of the two basic reaction chamber geometries shows an advantage of the radiation-cooled system, which resulted in a significant improvement of the CF 4 conversion compared to the previous system. In the case of the illustrated waste gas situation of 100 slm, the negative effect of decreasing efficiency caused by cooling power is not yet noticeable in the case of the falling film system, but is already clearly evident at 150 slm as previously shown [118]. In case of missing data within the table no test results were available. The CF 4 turnover as well as the concentration of occurring secondary metabolites were analytically evaluated by a combination of FTIR (detection of inert gases, CF 4 as well as short-chain organic transformation products) and GC-MS (detection of CF 4 and unknown organic transformation products).
To investigate the effect of the nozzle geometry on the plasma geometry, the geometry of the plasma discharge was directly analyzed at the separated plasma torch, and differences in operational parameters were evaluated by imaging techniques.
The radiation-cooled configuration of the predecessor model was convincing and was therefore used as the basic model for further development (Figure 17). Table 8. Efficiency overview of different systems-original system, optimized system and commercially available water film system (with water and nitrogen as plasma gases) at 100 slm as a function of CF 4 concentration at 20 kW of input power.  Figure 17. Prototype of the radiation cooled plasma system during start-up (left) and permanent operation (right).

Flow Simulation of the Reaction Chamber
Solving the problem of optimal injection and mixing of the waste gas into the plasma zone of the reaction chamber, flow and heat simulations with the simulation software Ansys Fluent 2018 were carried out using a system of a tube-in-tube reactor with a perforated inner wall. For the structural design of the reaction chamber, simulations with variation of the plasma gas flow, waste gas flow as well as the coupled electrical power were carried out. Exemplarily results of these simulations are presented in Figure 18a showing the original system of the radiation-cooled reactor at a power input of 10 kW, a waste air flow of 150 slm and a plasma gas flow of 30 slm, whereby the waste gas is injected directly at the torch head. The model shows a good mixing behavior due to an impingement jet of plasma gas and waste gas in the area of the entrance of the reaction chamber. However, a distinct temperature profile within the reactor also occurs, which indicates a resulting temperature of less than 2400 K shortly after injection of the waste gas. The minimum conversion temperature of CF4 (approx. 2500 K) described in the literature is thus already undercut in the reaction chamber head, so that a power input of 10 kW is too low for successful conversion of CF4 (corresponds to SIE = 1100 kWh•1000 m −3 ). In total, a temperature drop of approx. 600 K occurs over the length of the reaction chamber. Figure 18b,c shows the temperature and turbulence profiles for the tube-in-tube model at the same gas flows (150 slm of CF4-containing waste gas; 50 slm steam; 20 kW power input, corresponding to 2222 kWh•1000 m −3 ) injecting the waste gas via the perforations. The previously observed impact jet no longer occurs due to the lack of injection via the nozzles in the mixing plate between the burner and the reaction chamber. The simulation results in the perforation plane show for the temperature distribution a strong cooling plume in the area of the perforation, which continues upwards towards the reactor head. On the other hand, a strong temperature rise to approx. 2800-3000 K occurs just below the perforation over the entire cross-sectional area of the reactor. This is due to the principle of jet injection. The high exit velocity of the plasma gas initiates a suction effect for the air flowing in at the wall, which causes cool waste gas to be torn upwards. The waste gas is thus sucked into the center of the plasma flame. At the same time, the "negative" temperature profile leads to a strong radial outflow of the plasma gas towards the reactor wall below the injection zone, which can be further enhanced by using the radially expanding Laval nozzle. Since the temperature profile in the present case is very high overall, an efficient CF4 treatment seems possible. The gas velocities in the center of the reactor are very high and support explanations given above. The

Flow Simulation of the Reaction Chamber
Solving the problem of optimal injection and mixing of the waste gas into the plasma zone of the reaction chamber, flow and heat simulations with the simulation software Ansys Fluent 2018 were carried out using a system of a tube-in-tube reactor with a perforated inner wall. For the structural design of the reaction chamber, simulations with variation of the plasma gas flow, waste gas flow as well as the coupled electrical power were carried out. Exemplarily results of these simulations are presented in Figure 18a showing the original system of the radiation-cooled reactor at a power input of 10 kW, a waste air flow of 150 slm and a plasma gas flow of 30 slm, whereby the waste gas is injected directly at the torch head. The model shows a good mixing behavior due to an impingement jet of plasma gas and waste gas in the area of the entrance of the reaction chamber. However, a distinct temperature profile within the reactor also occurs, which indicates a resulting temperature of less than 2400 K shortly after injection of the waste gas. The minimum conversion temperature of CF 4 (approx. 2500 K) described in the literature is thus already undercut in the reaction chamber head, so that a power input of 10 kW is too low for successful conversion of CF 4 (corresponds to SIE = 1100 kWh·1000 m −3 ). In total, a temperature drop of approx. 600 K occurs over the length of the reaction chamber. Figure 18b,c shows the temperature and turbulence profiles for the tube-in-tube model at the same gas flows (150 slm of CF 4 -containing waste gas; 50 slm steam; 20 kW power input, corresponding to 2222 kWh·1000 m −3 ) injecting the waste gas via the perforations. The previously observed impact jet no longer occurs due to the lack of injection via the nozzles in the mixing plate between the burner and the reaction chamber. The simulation results in the perforation plane show for the temperature distribution a strong cooling plume in the area of the perforation, which continues upwards towards the reactor head. On the other hand, a strong temperature rise to approx. 2800-3000 K occurs just below the perforation over the entire cross-sectional area of the reactor. This is due to the principle of jet injection. The high exit velocity of the plasma gas initiates a suction effect for the air flowing in at the wall, which causes cool waste gas to be torn upwards. The waste gas is thus sucked into the center of the plasma flame. At the same time, the "negative" temperature profile leads to a strong radial outflow of the plasma gas towards the reactor wall below the injection zone, which can be further enhanced by using the radially expanding Laval nozzle. Since the temperature profile in the present case is very high overall, an efficient CF 4 treatment seems possible. The gas velocities in the center of the reactor are very high and support explanations given above. The described velocity profile is also shown in Figure 18c. The simulation results indicate that due to the reduced radiation cooling of the reactor outer wall, the gas temperature inside the combustion chamber can be kept above 2300 K even at gas flows of 300 slm (Figure 18d). Sustainability 2020, 12, x FOR PEER REVIEW 28 of 39 described velocity profile is also shown in Figure 18c. The simulation results indicate that due to the reduced radiation cooling of the reactor outer wall, the gas temperature inside the combustion chamber can be kept above 2300 K even at gas flows of 300 slm (Figure 18d).  This improved temperature profile within the reaction chamber could be confirmed in practical tests on the prototype (see Table 9). It can be seen that the prototype achieved higher efficiency than the predecessor model across all concentrations. In comparison to the commercial model, the efficiencies are significantly higher, too, with almost complete turnover of CF 4 . Table 9. Comparison of the transformation efficiency of CF 4 in the previous radiation-cooled combustion chamber compared to the innovation with integrated heat recovery for volume flows of 150 slm and 300 slm of waste gas with concentrations of 2790-22,000 ppm of CF 4 at a power input of 20 kW.

Ecological Evaluation of Thermal Plasmas
The present analytical data show a very high demand on electrical energy (SIE = 1100-2200 kWh·1000 m −3 ) compared to other waste gas treatment processes, which initially gives rise to strong doubts about the sustainability of these processes.
However, due to the high chemical stability, the high vapor pressure and the insolubility of CF 4 in water, conventional waste gas treatment processes are not applicable. Only post-combustion processes are used as an alternative, but these are characterized by a very low efficiency of 16-52% [143,144]. Due to this low efficiency and the high global warming potential of CF 4 (factor 7349 compared to CO 2 ), this impression is completely put into perspective, as shown in Table 10.
When treating emissions with a high global warming potential, it is therefore essential to achieve very high cleaning efficiencies, since the CO 2 fingerprint of these components is many times higher than the CO 2 emissions from energy generation. Nevertheless, energy-efficient methods represent a clear competitive advantage, especially in this application. Table 10. Balancing of emissions of CO 2 , CH 4 and N 2 O as greenhouse gases for combustion and NTP processes for treatment of a gas flow of 1000 m 3 ·h −1 for 8000 h·a −1 . Data based on CO 2 emissions of the German energy mix in 2019 (0.401 kg CO 2 ·kWh −1 ). * Analytical data of the authors. ** Experience values for combustion. *** Calculated value based on a concentration of 10 mg·L −1 of ammonia as fertilizer in the swamp phase at pH 7.8 and ammonium/ammonia distribution of Kollbach [145].

Conclusions and Perspectives
The present results show that both non-thermal and thermal plasmas can be used in a wide range of applications in waste gas treatment. Nevertheless, their application is subject to a number of obstacles that need to be resolved first.
(1) Previous investigations were almost exclusively carried out at laboratory scale. Reliable scale-up factors for the transfer into industrial-scale systems are almost completely missing. The applied laboratory-scale leads to very high SIE values, so that an economic estimation is mostly not possible. (2) Most studies deal with the treatment of individual components, but treatment of VOC mixtures is mainly not addressed. Removal efficiencies of single compounds are probably not transferable to situation in VOC mixtures, since VOC compounds compete for reactive plasma species, on the one hand, and, on the other hand, reactive interactions between different pollutants may occur. (3) The effect of interfering compounds in real waste gases like water saturated or oversaturated conditions, aerosols, dust and salt particles have not been sufficiently investigated. (4) Methods to reduce the energy demand of plasma systems like intermittent plasmas, heat recovery (only thermal plasma) or process combinations with low-energy process like biological approaches have hardly been investigated so far. (5) There is a clear lack of experiences about thermal (only thermal plasma) and chemical long-term stability of the materials used. Accordingly, the maintenance intervals are currently quite short. (6) Recently developed catalysts are mostly nanoparticles not suitable for waste gas applications.
Furthermore, their production process is complex, and therefore, production and application are not possible under economic aspects.
According to these restrictions, plasma processes are currently only used frequently in niches, such as treatment of CFCs or clean room air treatment. However, the research results presented here impressively underline the possibility of the large-scale application of these processes. Thus, there is a need to catch up on these six aspects before the objective of industrial-scale application is achieved, especially since these processes appear to be very interesting compared to conventional waste air treatment processes under economic and ecological aspects.