Generation of ozone by pulsed corona discharge over ... - CiteSeerX

INSTITUTE OF PHYSICS PUBLISHING

JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 38 (2005) 409–416

doi:10.1088/0022-3727/38/3/010

Generation of ozone by pulsed corona discharge over water surface in hybrid gas–liquid electrical discharge reactor Petr Lukes1,3 , Martin Clupek1 , Vaclav Babicky1 , Vaclav Janda2 and Pavel Sunka1 1

Department of Pulse Plasma Systems, Institute of Plasma Physics, Academy of Sciences of the Czech Republic, Za Slovankou 3, PO Box 17, 182 21 Prague 8, Czech Republic 2 Department of Water Technology and Environmental Engineering, Institute of Chemical Technology, Technicka 5, 160 28 Prague 6, Czech Republic E-mail: [email protected]

Received 5 October 2004, in final form 10 November 2004 Published 20 January 2005 Online at stacks.iop.org/JPhysD/38/409 Abstract Ozone formation by a pulse positive corona discharge generated in the gas phase between a planar high voltage electrode made from reticulated vitreous carbon and a water surface with an immersed ground stainless steel plate electrode was investigated under various operating conditions. The effects of gas flow rate (0.5–3 litre min−1 ), discharge gap spacing (2.5–10 mm), applied input power (2–45 W) and gas composition (oxygen containing argon or nitrogen) on ozone production were determined. Ozone concentration increased with increasing power input and with increasing discharge gap. The production of ozone was significantly affected by the presence of water vapour formed through vaporization of water at the gas–liquid interface by the action of the gas phase discharge. The highest energy efficiency for ozone production was obtained using high voltage pulses of approximately 150 ns duration in Ar/O2 mixtures with the maximum efficiency (energy yield) of 23 g kW h−1 for 40% argon content.

1. Introduction Electrical treatment of water by applying pulsed high voltage discharges directly in the aqueous solution has been used recently to degrade a variety of organic compounds [1–11], for sterilization [12–18] and also for modification of surface properties of polymeric materials [19]. It has been demonstrated that non-thermal plasmas generated by these discharges initiate a variety of chemical and physical effects in water including a high electric field, intense ultraviolet radiation, overpressure shock waves and, of particular importance, formation of various chemically active species (OH·, H· and O· radicals, H2 O2 ) [20–23]. It has been shown that the principle reactive species involved in degradation of organic compounds are the hydroxyl radical and hydrogen peroxide. Consequently, a number of studies have utilized gas phase discharges generated over a water surface 3

Author to whom any correspondence should be addressed.

0022-3727/05/030409+08$30.00

© 2005 IOP Publishing Ltd

to produce in the gas or at the gas–liquid interface strongly oxidative species, such as OH· radicals, O atoms and their reaction products (O3 , H2 O2 ), that can dissolve into the water to initiate oxidation processes [24–30]. The recently developed hybrid electrical discharge reactors combine gas phase non-thermal plasmas formed above the water surface with direct liquid phase corona-like discharges in the water. Two types of configurations have been constructed: the hybrid-parallel and hybrid- series [31, 32]. In the series configuration the high voltage point electrode is placed in the liquid phase, and the electrode, made from reticulated vitreous carbon (RVC), is placed in the gas phase above the gas–liquid interface. The parallel configuration employs a high voltage RVC electrode in the gas phase and a high voltage needle-point electrode in the liquid phase, with the ground electrode placed at the gas–liquid interface. The main advantage of these reactors is that they lead to the production of the same chemical and physical effects as initiated in the

Printed in the UK

409

P Lukes et al

individual gas and liquid phase discharges. A combination of these factors can lead to enhancement of the overall efficiency of the corona discharge process for removal of pollutants from water. Previous experiments have demonstrated simultaneous formation of ozone in the gas phase and hydrogen peroxide and hydroxyl radicals in the liquid phase in gas–liquid hybrid reactors [31]. Significant enhancement of the degradation of phenol and nitrobenzene was observed in the hybrid reactors compared with single-liquid phase discharge reactors [32–34]. It was shown that the degradation mechanism in the hybrid discharge reactor includes simultaneous oxidation by the OH· radicals produced by the discharge directly in the liquid phase and oxidation by OH· radicals and ozone formed by the gas phase discharge above the liquid. In addition, at high pH values it was observed that a synergistic combination of both gas and liquid discharges occurs through a Peroxone process between ozone formed by the gas phase discharge and hydrogen peroxide produced by the liquid phase discharge, which resulted in additional formation of OH· radicals available for the oxidation process in water [34]. Further development of hybrid gas–liquid electrical discharge reactors is needed since in the previous work it was not possible to independently control the gas phase discharge above the water surface and the liquid phase discharge. In order to optimize the power delivered into the two discharges generated in the gas and the liquid phases and to effectively tune the power supply to the hybrid discharge reactor we have constructed a gas–liquid discharge reactor with separately charged electrodes in the gas and liquid phases using two pulse power supplies. This allows independent control of the gas and liquid phase discharges including separate variation of the input power, voltage and pulse repetition rate for each phase. In this work, only one aspect of the overall design of a hybrid gas– liquid discharge reactor is considered, namely the gas phase discharge over the water surface. The formation of gaseous ozone in this reactor is investigated in order to assess the role of the gas phase discharge without interference from the liquid phase discharge. The effects of discharge gap spacing, electrical power applied to the reactor, gas flow rate and gas phase composition (oxygen and O2 /Ar, O2 /N2 mixtures) on ozone production are discussed.

2. Experimental section A schematic diagram of the apparatus for generating pulsed electrical discharge simultaneously in the aqueous phase and in the gas phase above the water surface is shown in figure 1. The reactor is a closed box with Plexiglass walls with outer dimensions of 200 × 200 × 200 mm3 . It consists of a two separately charged high voltage electrodes: one placed in the water and one placed in the gas phase above the water. The electrodes are separated by a circular perforated plate made from stainless steel connected with an electrically grounded stainless steel tube of inner diameter 160 mm. The high voltage needle electrode in the liquid is made from a mechanically sharpened tungsten wire placed along the axis of the stainless steel tubular ground electrode. The gas phase high voltage electrode is made from an RVC disc (diameter 50 mm × thickness 10 mm), which is attached to a stainless 410

Figure 1. Schematic of gas–liquid phase discharge reactor: 1, gas phase; 2, gas phase RVC electrode; 3, gas phase discharge; 4, ground electrode; 5, aqueous solution; 6, liquid phase point electrode; 7, liquid phase discharge; 8, reactor chamber; 9, insulator; HV, pulse power supply.

steel holder connected to the pulsed high voltage. The RVC electrodes were previously found to be well suited for gas phase pulse corona reactors since they allow production of uniform discharges on sharp edges of RVC, which serves as a very fine multipin electrode [35]. The distance between the RVC electrode and the ground stainless steel plate submerged in the water is fixed at 40 mm. The inner walls of the stainless steel tube above the plate are insulated to prevent discharging along the water surface to the walls of the stainless steel tube. The gas phase discharge gap in this work is defined as the region between the RVC electrode and the water surface, and its height was varied through adjustment of the volume of aqueous solution used in the reactor. Separate charging of the liquid phase needle electrode and the gas phase RVC electrode is provided by two pulse power supplies, which are identical to the power supply used in previous work [9–11, 20]. In this work one power supply was used to generate only the gas phase discharge above the aqueous solution, i.e. no discharge was generated in the liquid phase. It consists of a high voltage 0–50 kV dc source, a rotating double spark gap giving a maximum pulse repetition frequency of 100 Hz and a storage capacitor of 2 nF. The design permits variation of the applied voltage, its pulse repetition frequency and the discharge power and energy delivered into the discharge. In this work a pulsed high voltage of positive polarity in the range 15–30 kV with a fixed pulse repetition frequency of 50 Hz was used. The temporal evolution of the gap voltage and electrical current through the reactor were measured using a capacitive voltage divider and a Rogowski coil, respectively. The gap voltage and the total discharge current were recorded using a four-channel Hewlett Packard HP 54542 digital oscilloscope. Figure 2 shows typical waveforms of the pulse voltage and

Generation of ozone by corona discharge

Figure 2. Typical waveforms of pulse voltage and discharge current for pulsed corona discharge in oxygen. U = 20 kV, C = 2 nF, discharge gap spacing 5 mm.

pulse energy, Ep , was evaluated as the storage energy of the charged capacitor Ep = 21 CU 2 . The difference between the values of Ep determined as Ep = 21 CU 2 and Ep derived from the voltage and current waveforms was less than 10% when a storage capacitor of 2 nF was used. However, when the storage capacitor of 0.2 nF was used it was not possible to evaluate the Ep correctly from the voltage and current waveforms due to their oscillation characteristics (figure 3). Therefore, the Ep value determined as the storage energy of the charged capacitor was used in each case. Water was circulated through the reactor using a peristaltic pump with a flow rate of 0.4 litre min−1 and cooled to maintain a constant temperature in the reactor vessel of 18˚C. A solution of NaH2 PO4 (7.7 mmol l−1 ) with an initial conductivity of 500 µS cm−1 was used in each case. The solution conductivity remained constant during each experiment. Gas (oxygen, and O2 /N2 and O2 /Ar mixtures) flowed continuously through the reactor at atmospheric pressure and ambient temperature. The gas was fed into the reactor via an inlet port at the reactor wall. The concentration of ozone in the output gas was determined by the iodometric method by running the reactor outlet stream through a gas wash bottle containing KI solution [36]. In the case of ozone measurements in O2 /N2 mixtures, sodium azide (NaN3 ) was added to the KI solutions after bubbling with output gas and before acidification by sulfuric acid to eliminate interferences caused by nitrites in the iodometric method. Nitrites can be formed in the KI solution from nitrogen oxides produced by the discharge and subsequently dissolved from the gas phase into the KI solution (equation (1)). NO(g) + NO2 (g) + H2 O → 2H+ + 2NO− 2.

(1)

Without addition of azide, nitrites initiate, after acidification of KI solution, a chain reaction with iodide and dissolved oxygen, which makes it impossible to reach the endpoint of the iodometric titration (equations (2) and (3)). − + 2NO− 2 + 2I + 4H → 2NO + I2 + 2H2 O,

(2)

− 2NO + O2 + H2 O → 2H+ + NO− 2 + NO3 .

(3)

It should be noted that the exact order of azide addition is very important since azide otherwise rapidly decomposes ozone (equation (4)). Figure 3. Typical waveforms of pulse voltage and discharge current for pulsed corona discharge in oxygen. U = 20 kV, C = 0.2 nF, discharge gap spacing 5 mm.

current for gas corona discharges in oxygen using a charging capacitance of 2.0 nF and a discharge gap of 5 mm. As can be seen, the total current flowing through the reactor consists of two parts: the capacitive current, with a rise time of about 20 ns and peak value of 130 A caused by charge released by the capacitor, and the corona current of a pulse duration of about 500 ns full-width at half-maximum (FWHM) and a peak value of about 55 A. Figure 3 shows the temporal evolution of voltage and current for a gas phase corona discharge generated in an oxygen atmosphere using a high voltage cable (l = 2 m, Z = 50
) as the storage capacitor (C = 0.2 nF). The electrical power, P , applied to the reactor was calculated from the applied voltage, U , charging capacity, C, and pulse repetition frequency, f , as P = f Ep , where the

O3 + 2N3− + 2H2 O → O2 + 3N2 + 2OH− .

(4)

By using azide in the modified procedure of the iodometric method described above, nitrites were effectively removed from the solution (equation (5)) [37]. + 3N3− + NO− 2 + 4H → 5N2 + 2H2 O.

(5)

The presence of azide did not interfere with the iodometric method. After reaching the titration endpoint the excess of azide was destroyed by addition of NaNO2 to the KI solution.

3. Results and discussion 3.1. Effect of gas flow rate on ozone formation Figure 4 shows the effect of gas flow rate on (a) ozone concentration and (b) ozone production efficiency (also termed 411

P Lukes et al

Figure 5. Ozone concentration as a function of specific energy density for various oxygen flow rates: ◦, 0.5 litre min−1 ;
, 1.0 litre min−1 ;  , 2.0 litre min−1 ; , 3.0 litre min−1 . Discharge gap spacing 5 mm, C = 2 nF.

Figure 4. (a) Ozone concentration and (b) ozone production efficiency measured in oxygen as a function of gas flow rate for different power inputs applied to the reactor. Discharge gap spacing 5 mm, C = 2 nF. Power input: ◦, 11 W;
, 22 W;  , 36 W.

energy yield) measured in oxygen for different power inputs applied to the reactor at a fixed discharge gap spacing of 5 mm. The power input was varied from 11 to 36 W using different applied voltages from 15 to 27 kV at a fixed pulse repetition frequency of 50 Hz. The gas flow rate was varied from 0.5 to 3.0 litre min−1 at each applied power input. Reported ozone concentrations are steady state values of the volume fractions of ozone in the output gas, in terms of parts per million (ppmv). Figure 4(a) shows that the ozone concentration increased with increasing power input and decreased with increasing oxygen flow rate. This is expected because lower flow rates give longer residence times where the gas is subjected to the discharge as well as higher power input delivered into the discharge. Both effects lead to larger ozone formation. On the other hand, the opposite effect was observed for ozone production efficiency (figure 4(b)), which decreased with increasing power input and, less significantly, with lower flow rates ( 0.5 Wh l−1 . Therefore, to minimize thermal decomposition of ozone, flow rates less than 1.5 litre min−1 seem to be optimal for the range of power inputs used in this work (up to 45 W). Figure 4(b) also shows constant ozone production efficiencies for flow rates above 1 litre min−1 at each applied power input. 3.2. Effect of discharge gap spacing on ozone formation Figure 6 shows the dependence of (a) ozone concentration and (b) ozone production efficiency on the applied power input for different discharge gap spaces between the high voltage RVC electrode and the water surface measured in oxygen. The power input was varied from 11 to 45 W using different applied voltages from 15 to 30 kV at a constant pulse repetition frequency of 50 Hz. The discharge gap space was varied from 2.5 to 10 mm at each applied power input. It is apparent that for the same power input the ozone concentration and production efficiency increase with increasing gap spacing. However, when the same discharge gap was used, the ozone concentration increased with increasing power input, in contrast to the effect observed for ozone energy yield. For a fixed pulse repetition rate and discharge gap spacing, d, it can be expected that a higher power input results in a larger average electric field (E = U/d), and higher average power density in the discharge (P /Sd), where S is the surface area of the RVC electrode. Increasing the electric field implies

Generation of ozone by corona discharge

Figure 7. Ozone concentration measured in Ar/O2 and N2 /O2 mixtures as a function of oxygen concentration. Total gas flow rate 2.5 litre min−1 , U = 20 kV, C = 2 nF, P = 20 W, discharge gap 5 mm. Table 1. Relative ratio, XO3 , between ozone production obtained in N2 /O2 and Ar/O2 mixtures and pure oxygen for an applied voltage of 20 kV and charging capacitances of 2 nF (P = 20 W) and 0.2 nF (P = 2 W). Pulse repetition frequency, 50 Hz; discharge gap spacing, 5 mm; total gas flow rate, 2.5 litre min−1 . N2 /O2

Figure 6. (a) Ozone concentration and (b) ozone production efficiency measured in oxygen as a function of applied power input for different discharge gap spacing. Gas flow rate 2.5 litre min−1 , C = 2 nF. Gap spacing: •, 2.5 mm;  , 5 mm;
, 7.5 mm; , 10 mm.

increasing E/N and a higher mean electron energy in the discharge. Therefore, with a higher power input a larger amount of atomic oxygen can be formed through electron impact dissociation of oxygen, which results in a larger generation of ozone [41, 42]. At the same time, however, a decrease in ozone energy yield with higher applied power input indicates that increases in power density in the discharge can lead to higher power consumption by heat dissipation and subsequently to a larger thermal decomposition of ozone in the discharge. In addition to the effects of gap spacing on the electric field and power density in the discharge, a larger discharge gap leads to formation of a significantly different kind of discharge. In small gaps (
5 mm) very thin glow-like corona discharge filaments, which homogenously filled the discharge gap above the water surface, were formed during each pulse. With larger discharge gaps smaller numbers of brighter tuftlike corona discharge filaments were formed in each pulse, and in the discharge gap of 10 mm only single bright discharge channels were formed per pulse. Therefore, as the number of plasma channels decreases with increasing discharge gap, the volume of plasma decreases and the charge transferred in each

Ar/O2

(%O2 )

2 nF

0.2 nF

2 nF

0.2 nF

0.2 0.4 0.6 0.8 1.0

0.2 0.1 0.1 0.3 1.0

0.5 0.7 0.9 1.0 1.0

1.4 1.5 1.2 1.1 1.0

1.8 1.6 1.4 1.2 1.0

plasma channel increases for the same applied voltage. This implies a higher electron mean energy and gas temperature in the discharge, which could explain the increase in ozone production rate with decreasing energy efficiency observed with larger discharge gap spaces. 3.3. Effect of gas composition on ozone formation Figure 7 shows the effects of gas composition on ozone production measured in N2 /O2 and Ar/O2 mixtures for different oxygen content. The same total gas flow rate of 2.5 litre min−1 , power input of 20 W (U = 20 kV, f = 50 Hz, Ep = 0.4 J pulse−1 ) and discharge gap spacing of 5 mm were used in each case. Consequently, table 1 shows the relative ratio, XO3 , between the ozone concentration obtained in Ar/O2 and N2 /O2 mixtures, (cO3 )mix , and the theoretical value of ozone obtained in pure oxygen, (cO3 )O2 , at a correspondingly reduced O2 percentage in the gas (%O2 ), calculated as [39] XO3 =

(cO3 )mix . (cO3 )O2 × (%O2 )

(6)

It is apparent that the presence of argon and nitrogen in the mixtures with oxygen had different effects on the production of ozone. In Ar/O2 mixtures the ozone formation decreased with increasing argon content (figure 7); however, >1 value of the XO3 obtained for all Ar/O2 mixtures indicates improved ozone production in the presence of argon. On the other hand, 413

P Lukes et al

in N2 /O2 mixtures the ozone production was considerably suppressed by the presence of nitrogen (figure 7). It is known that both nitrogen and argon can have catalytic effects on the production of ozone [38, 42–45]. In the case of nitrogen such an effect is caused by the reactions of nitrogen atoms and electronically excited nitrogen molecules, N2 (A 3 u+ ) and N2 (B 3 g ), with oxygen (equations (7)–(10)), which can produce additional oxygen atoms for generation of ozone. (7) N· + O2 → NO + O·, N· + NO → N2 + O·,

(8)

N2 (A) + O2 → N2 O + O·,

(9)

N2 (A, B) + O2 → N2 + 2O·.

(10)

However, at higher specific energy densities in the discharge such a positive effect of nitrogen can be counteracted by quenching of oxygen atoms and destruction of ozone by the relatively high concentrations of nitrogen atoms and nitrogen oxides (equations (11)–(14)) [38, 46–48]. O· + NO2 → NO + O2 ,

(11)

O· + NO + N2 → NO2 + N2 ,

(12)

N· + O3 → NO + O2 ,

(13)

NO + O3 → NO2 + O2 .

(14)

In Ar/O2 mixtures these side reactions cannot occur since argon as an monoatomic and chemically inert gas cannot serve as a chemical quencher of atomic oxygen and its catalytic effect is mostly due to the involvement of argon as the third collision partner, M, in the ozone formation (equation (15)): O· + O2 + M → O3 + M

(M = O2 , N2 , Ar).

(15)

Therefore, the significantly reduced production of ozone in the N2 /O2 mixtures shown in figure 7 can be explained by the inhibition effect of N atoms and NOx (equations (11)–(14)) caused most likely by the excessive energy density in the discharge. Since β depends for a given flow rate directly on the power input, the improvement in the production of ozone in the N2 /O2 mixtures should be obtained primarily by reduction of the applied power in the reactor through reduction of the input pulse energy and/or pulse repetition rate. However, the same effect of nitrogen on ozone formation was also observed for lower pulse energies (∼0.2 J pulse−1 ) and lower power inputs (11 W) applied in the discharge. These values are actually even lower than those used in previous work for generation of ozone in air by pulsed positive corona discharge in a reactor of coaxial wire–cylinder geometry (Ep = 1.8 J pulse−1 , P = 18 W) [49]. In this case, however, voltage and current pulses of duration 100–150 ns FWHM were delivered into the discharge, compared with typical values of about 500 ns width of voltage and current pulses generated in this work (figure 2). A high voltage of similar pulse width (∼100–200 ns) was also used by other authors for ozone generation by pulsed corona discharges in air [50–52]. Consequently, Chalmers et al [53] reported of the ozone yield that were strongly dependent on the pulse width using corona discharges in oxygen. Therefore, the pulse duration of applied power in the discharge has also to be considered since with 414

Figure 8. Ozone concentration measured in Ar/O2 and N2 /O2 mixtures as a function of oxygen concentration. Total gas flow rate 2.5 litre min−1 , U = 20 kV, C = 0.2 nF, P = 2 W, discharge gap 5 mm.

longer pulses a larger fraction can be lost in processes which do not contribute to the formation of ozone and result in heat dissipation [54]. To decrease both the magnitude and pulse duration of the power delivered into the discharge, a charging capacitance of 0.2 nF was used instead of 2.0 nF. This lower capacitance was obtained by utilizing a high voltage cable (l = 2 m, Z = 50
) as a storage capacitor. This resulted in the formation of voltage and current pulses with oscillatory characteristics and with a pulse duration of approximately 150 ns FWHM (figure 3). In addition, a glow-like corona discharge was formed even at higher discharge gaps (>10 mm) compared with the formation of single plasma channels using a capacitance of 2 nF (see section 3.2.). Figure 8 shows ozone formation in N2 /O2 and Ar/O2 mixtures using a charging capacitance of 0.2 nF. The same applied voltage of 20 kV (P = 2.0 W, Ep = 0.04 J pulse−1 ), gas flow rate of 2.5 litre min−1 and discharge gap of 5 mm were used as in the experiments with a charging capacitance of 2 nF (figure 7). Comparison of the trends in ozone formation in figures 7 and 8 and the relative ratios, XO3 , in table 1 indicate an improvement in the ozone production efficiency using a smaller charging capacitance, especially in N2 /O2 mixtures, although the total amount of ozone formation was smaller than for the 2 nF case. Higher concentrations of ozone and significant increases in ozone production efficiency were observed with further increases in the applied voltage in the N2 /O2 and Ar/O2 mixtures. Moreover, in Ar/O2 mixtures ozone was formed in larger amounts than obtained in pure oxygen. Figure 9 shows the ozone formation in N2 /O2 and Ar/O2 mixtures for an applied voltage of 30 kV with a discharge gap of 5 mm (P = 4.5 W, Ep = 0.09 J pulse−1 ). This figure shows that the production of ozone in Ar/O2 mixtures with argon content in the range 10–70% was significantly enhanced, with the maximum amount of ozone production occurring with 40% argon. This corresponded to the maximum ozone production efficiency of 23 g kW h−1 . A similar effect of argon on ozone formation was also observed by Manning [44, 45]. It is not

Generation of ozone by corona discharge

discharge above the water surface. Ono and Oda [52] observed that the addition of 2.4 vol% water vapour to dry air reduced ozone production by pulsed corona discharges by a factor of 6. The water vapour can absorb a substantial part of the electron energy of the discharge that could otherwise be used in the ozone formation process. The collision of water molecules with electrons from the gas phase discharge can proceed via electron dissociation of water to form H· and OH· radicals that subsequently destroy ozone molecules (equations (16)–(18)).

Figure 9. Ozone concentration measured in Ar/O2 and N2 /O2 mixtures as a function of oxygen concentration. Total gas flow rate 2.5 litre min−1 , U = 30 kV, C = 0.2 nF, P = 4.5 W, discharge gap 5 mm.

the aim of this work to discuss fully the mechanism of ozone formation in Ar/O2 mixtures; however, it was suggested in [45, 55] that apart from the three body reaction (equation (15)), argon may affect the efficiency of ozone production also due to other factors (e.g. higher ionization potential and lower specific heat capacity of argon compared with oxygen, electronegativity of oxygen compared with electron nonattaching properties of argon, etc), which leads to the formation of discharges of higher average electron density in a Ar/O2 mixtures when compared with pure oxygen and, thus, to increased ozone formation. At the same time, the energetic or excited molecular species of oxygen contain more energy in rotational, vibrational and translational modes and lose energy at a much slower rate compared with monoatomic argon, which causes a decrease in the number of destructive energetic collisions with ozone in Ar/O2 mixtures when compared with a pure oxygen [45]. 3.4. Effect of humidity on ozone formation The efficiency of ozone production was significantly enhanced when a high voltage of shorter pulse width was applied to the discharge. However, the ozone yields (∼10–20 g kW h−1 ) are still much lower than the yields reported for ozone production in other work using corona discharges in dry oxygen or air (∼50–150 g kW h−1 ) [44, 49, 50, 56]. Such a discrepancy may be due to several reasons. One of the reasons is the loss of energy applied to the discharge in Joule heating of the water layer above the immersed ground electrode. Therefore, the energy consumption of the ozone formation by the gas phase discharge includes to some extent also these energy losses. Another reason is the presence of water vapour in the discharge gap between the high voltage RVC electrode and the water surface. Clearly, since plasma channels formed in the gas phase are in direct contact with the water surface, the solution can be locally heated and vaporized by the action of the discharge. Typically, the concentrations of water measured in the output gas were 1.5–2.0 vol% compared with less than 0.1 vol% of water in the inlet gas, but it is possible that even higher concentrations occur in the region of the gas phase

H· + O3 → OH· + O2 ,

(16)

OH· + O3 → HO2 · + O2 ,

(17)

HO2 · + O3 → OH· + 2O2 .

(18)

In addition, due to the much larger quenching rate of O(1 D) by H2 O than by N2 , O2 or Ar (about 6–7 times higher) [57], oxygen atoms formed by electron impact dissociation of oxygen can also combine with water molecules to produce OH· radicals, preventing ozone formation (equation (19)). O(1 D) + H2 O → OH· + OH·.

(19)

In the presence of nitrogen the electronically excited nitrogen molecules can dissociate H2 O to form OH· and H· radicals (equation (20)) [46–48, 58, 59]. N2 (A) + H2 O → N2 + OH· + H·.

(20)

Therefore, it is apparent that water vapour can significantly affect the ozone formation by the gas phase discharge generated above the water surface. However, it should be noted that formation of OH· radicals by the gas phase discharge is also desired since these radicals can significantly contribute to the degradation of aqueous organic compounds at the gas– liquid interface as was shown in previous work [24, 29, 34]. Thus, a compromise between production of ozone in high concentrations and with high energy efficiency, and production of OH· radicals by the gas phase discharge generated above the aqueous solution has to be considered in further development of hybrid gas–liquid discharge reactors. It should also be emphasized that rather than development of a new type of ozone generator the final objective is determination of conditions that will result in the most efficient utilization of chemically active species (O3 , OH· and O radicals, and other species) formed by the gas phase discharge in the removal of pollutants from water. Such conditions will be elucidated in further work.

4. Conclusions The effects of discharge gap spacing, electrical power applied to the reactor and gas phase composition (oxygen and Ar/O2 , N2 /O2 mixtures) on the production of ozone generated by the gas phase discharge above an aqueous solution in a hybrid gas–liquid phase electrical discharge reactor were investigated. It was shown that ozone production increases with increasing input power and larger discharge gaps. The pulse duration of the power applied into the discharge had a significant effect on the production of ozone, especially in N2 /O2 mixtures. Improvement in the ozone production 415

P Lukes et al

efficiency was obtained by applying high voltage pulses of approximately 150 ns duration compared with 500 ns in Ar/O2 mixtures (23 g kW h−1 ). The production of ozone was strongly affected by the presence of water vapour formed through vaporization of water at the surface by the action of the gas phase discharge.

Acknowledgment This work was supported by the Grant Agency of the Czech Republic under contract No 202/02/1026.

References [1] Malik M A, Ghaffar A and Malik S A 2001 Plasma Sources Sci. Technol. 10 82–91 [2] Akiyama H 2000 IEEE Trans. Dielectr. Electr. Insul. 7 646–53 [3] Sun B, Sato M and Clements J S 2000 Environ. Sci. Technol. 34 509–13 [4] Sharma A K, Locke B R, Arce P and Finney W C 1993 Hazard. Waste Hazard. Mater. 10 209–19 [5] Gao J, Liu Y, Yang W, Pu L, Yu J and Lu Q 2003 Plasma Sources Sci. Technol. 12 533–8 [6] Malik M A 2003 Plasma Sources Sci. Technol. 12 S26–32 [7] Willberg D M, Lang P S, Hochemer R H, Kratel A and Hoffmann M R 1996 Environ. Sci. Technol. 30 2526–34 [8] Grymonpre D, Sharma A K, Finney W C and Locke B R 2001 Chem. Eng. J. 82 189–207 [9] Sunka P, Babicky V, Clupek M, Lukes P, Simek M, Schmidt J and Cernak M 1999 Plasma Sources Sci. Technol. 8 258–65 [10] Lukes P, Clupek M, Sunka P, Babicky V and Janda V 2002 Czech. J. Phys. 52 D800–6 [11] Lukes P, Clupek M, Babicky V, Sunka P, Winterova G and Janda V 2003 Acta Phys. Slovaca 53 423–8 [12] Mizuno A and Hori Y 1988 IEEE Trans. Indust. Appl. 24 387–94 [13] Sato M, Ohgiyama T and Clements J S 1996 IEEE Trans. Indust. Appl. 32 106–12 [14] Ching W K, Colussi A J, Sun H J, Nealson K H and Hoffmann M R 2001 Environ. Sci. Technol. 35 4139–44 [15] Schmelev V M, Evtyukhin N V and Che D O 1996 Chem. Phys. Rep. 15 463–8 [16] Abou-Ghazala A, Katsuki S, Schoenbach K H, Dobbs F C and Moreira K R 2002 IEEE Trans. Plasma Sci. 30 1449–53 [17] Anpilov A M, Barkhudarov E M, Christofi N, Kopev V A, Kossyi I A, Taktakishvili M I and Zadiraka Y 2002 Lett. Appl. Microbiol. 35 90–4 [18] van Heesch E J M, Pemen A J M, Huijbrechts P A H J, van der Laan P C T, Ptasinski K J, Zanstra G J and de Jong P 2000 IEEE Trans. Plasma. Sci. 28 137–43 [19] Brablec A, Slavicek P, Stahel P, Cizmar T, Trunec D, Simor M and Cernak M 2002 Czech. J. Phys. 52 D491–500 [20] Sunka P 2001 Phys. Plasmas 8 2587–94 [21] Robinson J W, Ham M and Balaster A N 1973 J. Appl. Phys. 44 72–5 [22] Joshi A A, Locke B R, Arce P and Finney W C 1995 J. Hazard. Mater. 41 3–30 [23] Sun B, Sato M and Clements J S 1997 J. Electrostat. 39 189–202 [24] Hoeben W F L M, van Veldhuizen E M, Rutgers W R, Cramers C A M G and Kroesen G M W 2000 Plasma Sources Sci. Technol. 9 361–9

416

[25] Sharma A K, Camaioni D M, Josephson G B, Goheen S C and Mong G M 1997 J. Adv. Oxid. Technol. 2 239–47 [26] Anpilov A M et al 2001 J. Phys. D: Appl. Phys. 34 993–9 [27] Velikonja J, Bergougnou M A, Castle G S P, Cairns W L and Inculet I I 2001 Ozone Sci. Eng. 23 467–78 [28] Piskarev I M, Rylova A E and Sevast’yanov A I 1996 Russ. J. Electrochem. 32 827–9 [29] Hayashi D, Hoeben W F L M, Dooms G, van Veldhuizen E M, Rutgers W R and Kroesen G M W 2000 J. Phys. D: Appl. Phys. 33 2769–74 [30] Sano N, Kawashima T, Fujikawa J, Fujimoto T, Kitai T and Kanki T 2002 Indust. Eng. Chem. Res. 41 5906–11 [31] Lukes P, Appleton A T and Locke B R 2004 IEEE Trans. Indust. Appl. 40 60–7 [32] Grymonpre D R, Finney W C, Clark R J and Locke B R 2004 Ind. Eng. Chem. Res. 43 1975–89 [33] Appleton A T, Lukes P, Finney W C and Locke B R 2002 Contributed Papers HAKONE VIII (P¨uhaj¨arve, Estonia) pp 313–7 [34] Lukes P, Kirkpatrick M J, Finney W C and Locke B R 2004 Proc. 4th Int. Symp. Non Thermal Plasma Technol. (Panama City Beach, USA) pp 120–5 [35] Kirkpatrick M, Finney W C and Locke B R 2000 IEEE Trans. Indust. Appl. 36 500–9 [36] Clesceri L S, Greenberg A E and Eaton A D 1998 Standard Methods for the Examination of Water and Wastewater (Washington: APHA, AWWA and WEF) pp 2–43 [37] Alsterberg G 1925 Biochem. Z. 159 36–47 [38] Eliasson B and Kogelschatz U 1991 IEEE Trans. Plasma Sci. 19 309–23 [39] Held B and Peyrous R 1998 Eur. Phys. J. Astrophys. 4 73–86 [40] Peyrous R, Monge C and Held B 1999 Czech J. Phys. 49 289–99 [41] Eliasson B and Kogelschatz U 1986 J. Phys. B: At. Mol. Phys. 19 1241–7 [42] Penetrante B M, Bardsley J N and Hsiao M C 1997 Japan. J. Appl. Phys. 36 5007–17 [43] Mason N J, Skalny J D and Hadj-Ziane S 2002 Czech. J. Phys. 52 85–94 [44] Manning T J 2000 Ozone Sci. Eng. 22 53–64 [45] Manning T J and Hedden J 2001 Ozone Sci. Eng. 23 95–103 [46] Mok Y S and Ham S W 1998 Chem. Eng. Sci. 53 1667–78 [47] Chang J S, Lawless P A and Yamamoto T 1991 IEEE Trans. Plasma Sci. 19 1152–66 [48] Mukkavilli S, Lee C K, Varghese K and Tavlarides L L 1988 IEEE Trans. Plasma Sci. 16 652–60 [49] Simek M and Clupek M 2002 J. Phys. D: Appl. Phys. 35 1171–5 [50] Samaranayake W J M, Miyahara Y, Namihira T, Katsuki S, Sakugava T, Hackman R and Akiyama H 2000 IEEE Trans. Dielectr. Electr. Insul. 7 254–60 [51] Hegeler F and Akyiama H 1997 IEEE Trans. Plasma Sci. 25 1158–65 [52] Ono R and Oda T 2003 J. Appl. Phys. 93 5876–82 [53] Chalmers I D, Zanella L and MacGregor S J 1995 10th IEEE Int. Pulsed Power Conf., Digest of Technical Papers (Albuquerque, USA) vol 2, pp 1249–54 [54] Chang M B, Balbach J H, Rood M J and Kushner M J 1991 J. Appl. Phys. 69 4409–17 [55] Skalny J D, Mikoviny T, Mason N J and Sobek V 2002 Ozone Sci. Eng. 23 29–37 [56] Ahn H S, Hayashi N, Ihara S and Yamabe C 2003 Japan. J. Appl. Phys. 42 6578–83 [57] Falkenstein Z 1999 Ozone Sci. Eng. 21 583–603 [58] Peyrous R 1990 Ozone Sci. Eng. 12 19–40 [59] Peyrous R 1990 Ozone Sci. Eng. 12 41–64