Interface reactions between water and ionic wind generated by d.c. corona discharge in nitrogen

This work investigated the interface reactions between water and ionic wind from a positive d.c. corona discharge in nitrogen. The effects of drift ions and active neutral species were investigated on their production of hydrogen peroxide in water. Both positive drift ions and active neutral species were found to produce hydrogen peroxide in water. 1 mol positive drift ions can produce 0.13 mol hydrogen peroxide under atmospheric pressure; a reduced gas pressure increased the production of hydrogen peroxide, with 1 mol ions producing 0.29 mol hydrogen peroxide at 100 torr. The gas pressure has little effect on the production of hydrogen peroxide by active neutral species. The drift positive ions acted as ion anode on water surface, reacting with water to produce hydrogen, oxygen, and hydrogen peroxide. The adsorbed form of OH and O produced at the interface can synergistically work to react with the OH scavengers in water. Both HO2 and H2O2 from the discharge were transported to water to produce the liquid phase hydrogen peroxide.


Introduction
The interactions between water and plasma have been studied for more than two hundred years since the first report by Cavendish's in 1785. In recent researches, the interfacial processes between plasma and liquid have been deeply investigated, which include the processes caused by neutral species, ions, electrons and plasma electrochemistry. At the plasma-liquid interface, neutral species such as O, O 3 , OH, NO, HO 2 , H 2 O 2 , HNO 2 and HNO 3 from the discharge can be transported to the liquid (Bruggemann et al 2016). * Author to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Sano investigated the interface reactions between water and corona discharges in N 2 /O 2 , and found that the neutral species like oxygen atom was generated in the presence of oxygen, and then carried by ionic wind to the water surface, producing aqueous hydroxyl radicals (Sano et al 2002).
The reactions of energetic positive ions at the plasma-liquid interface was also investigated. The atomic simulations have shown that 100 eV O 2 + can penetrate to 3 nm below the liquid surface (Nikiforov 2008). Water molecules can be sputtered to the gas phase after impinged by the energised ions (Minagawa et al 2014). Energised ions bombard the water surface to excite, ionize or dissociate the water molecules, and produce hydroxyl radicals, which have short lifetime of 2.7 μs at the interface (Attri et al 2015). The kinetic energy of positive ions bombarding the water cathode causes H 2 O excitation and ionization, and eventually dissociation to protons and hydroxyl radicals (Cserfalvi and Mezei 1996). The plasma induced electrochemical process at the interface has been proved able to exceed the 100% current efficiency by Faraday's law based upon the stoichiometry of the given electrochemical reactions. Hickling investigated contact glow discharge electrolysis, and found that the hydrogen peroxide produced in the electrolyte can reach 1.8 times the predicted by Faraday's law at 760 torr; the hydrogen peroxide formation was thought due to the positive ions accelerated through the cathode fall entering the liquid and dissociating water molecules (Hickling and Ingram 1964). Sengupta analysed the chemical yields of contact glow discharge electrolysis, and found that water molecule dissociation partly occurred in the plasma near anode to produce H 2 and O 2 , and in anolyte to produce H 2 , O 2 and H 2 O 2 (Sengupta and Singh 1994). Kanzaki believed that H 2 O + was produced in gas phase in anode glow discharge electrolysis (Kanzaki et al 1986). In plasma electrolysis, one electron can participate in multiple reactions in gas phase, which increased the current efficiency. The energised electrons can excite, ionize or dissociate water molecules at the interface; non-energised electrons transported to liquid eventually become solvated electrons, which can recombine with cations or react with water molecules to produce hydroxide ions and hydrogen. Locke et al reviewed the energy yield of hydrogen peroxide production for various plasma-water interface reactions; most results obtained were between 0-1 g kWh −1 (Locke and Shih 2011). The low energy yield limits the application of plasma-induced advanced oxidation processes for water purification.
For a water cathode glow discharge, the average kinetic energy of positive ions entering the water was suggested to be greater than 100 eV (Sengupta et al 1998). The mean kinetic energy of drift ions, produced by corona discharge, reaching the water surface is only 0.01-0.1 eV (Chen and Davidson 2003). There have been few investigations of the reactions between positive drift ions and liquid water. In the present study, the mechanisms of positive drift ions and active neutral species from the corona discharge in nitrogen were evaluated in their ability to produce hydrogen peroxide in water. The effects of gas pressure, flow rate and the interface reaction mechanisms were investigated.

Electrical circuit
The experimental layout is shown in figure 1(a). Corona discharges were generated using needle electrodes, charged by a Glassman, PS/EJ20R30 d.c. power supply. The current limiting resistors are R 1 and R 2 ; the current-viewing resistor is R 3 . R 2 has a large resistance of 24.9 MΩ, which ensures the current is d.c. without pulses and protect the probe on R 3 once pulse current occurred. The voltage applied to the needle electrode was measured by a Tektronix P6015A high voltage probe. The gap voltage from needle to cathode was calculated by subtracting the partial voltage across R 2 (which equals to 747 V at 30 μA d.c. current) from the needle voltage measured by the high voltage probe. The voltage on the CVR was measured by a Teledyne LeCroy PP008 probe. Both voltage waveforms were recorded using a Teledyne LeCroy Waverunner 610Zi oscilloscope.

Reactor design
The chamber of the gas-sealed reactor encompassed an acrylic tube, with an inner diameter of 100 mm and a height of 100 mm. The gas inlet and outlet ports were situated on the lid. The tip radius of the needles was 0.2 mm. Inside the reactor, either a water cathode or a stainless-steel mesh cathode was used in the experiment. For experiments with water cathode, positive corona discharges were generated between a nine-needle high-voltage (HV) electrode and the sample solution. The gap between the needle tips and the solution surface was fixed at 10 mm. The solution was grounded through a stainless-steel plate at the bottom of the container, as illustrated in figure 1(d).
For experiments with mesh cathode, positive corona discharge was generated between the nine-needle HV electrode and the grounded mesh. The mesh was placed 5.3 mm above the solution surface, covering the entire surface. The gap between the needle tips and the mesh was 10 mm. The positive ions were filtered by the mesh, allowing only neutral species to reach the solution. The mesh pore size was 1 mm, and the wire diameter was 0.3 mm. The thickness of the mesh was 0.6 mm.

Gas preparation
For each experiment, the reactor was vacuumed to lower than 4 torr and refilled with flowing nitrogen. The process was repeated three times. The gas pressure was measured and controlled using an upstream pressure controller Alicat Scientific PC-30PSIA-D/5P, with a range of 0-1500 torr absolute, a resolution of 1 torr, and accuracy of 0.25%. The gas flow rate was set within the range of 0-10 slpm using an Alicat Scientific MC-10SLPM-D/5M mass flow controller, with a resolution of 0.01 slpm, and accuracy of 0.8%. The flow mass controller was put at the downstream side of the reactor to coordinate with the upstream pressure controller to set the working gas pressure and flow rate.
For the tests using deionized water, 5 ml deionized water was transferred into the sample container, via a 5 ml pipette (Gilson, P5000G). For the tests using TB solution or DMSO solution, 4.5 ml of deionized water was transferred into the sample container. Then, 0.5 ml of 0.1 mol l −1 TB solution or DMSO solution was added to achieve a concentration of 0.01 mol l −1 .
The H 2 O 2 produced in the solution was measured immediately after the exposure to the ionic wind generated by the corona discharge. Firstly, 3 ml of treated solution was mixed with 0.3 ml of 0.1 mol l −1 Ti 4+ solution and 0.3 ml of 1 mol l −1 sulfuric acid solution. The peroxotitanium complex Ti(O 2 )OH(H 2 O) 3 was formed in the solution, exhibiting a yellow colour and peak absorption at 390 nm, measured using a Thermo Scientific Evolution 201 spectrophotometer. The absorption and hydrogen peroxide concentration measurements were calibrated against a reference solution of H 2 O 2 with known concentration.
To measure pH and conductivity, 3 ml of treated solution was mixed with 12 ml of deionized water immediately after each test. The solution pH was measured with a Thermo Scientific Orion Star A211 pH metre, and the conductivity was measured with a Thermo Scientific Orion Star A325 pH/conductivity metre. The results were converted to the predilution values.
The concentration of nitrite and nitrate in the solution was measured by a colorimetric test kit (Sigma-Aldrich, Cat. No. 11746081001). The nitrite reacts with sulfanilamide and N-(1-naphthyl)-ethylene-diamine dihydrochloride to form a diazo dye, exhibiting a red colour and peak absorbance at 540 nm which can be measured by a spectrophotometer. The treated samples were divided into two groups, in one group the nitrite is measured; in the other group, the nitrate was reduced to nitrite by nicotinamide adenine dinucleotide phosphate in the presence of the enzyme nitrate reductase, and the total nitrite was measured. By comparing the difference of nitrite concentration, the nitrate concentration can be determined. Three independent experiments were conducted for each test under the same experimental conditions and procedures to ensure the results are reproducible. The mean value was used as the data point on the graph with the uncertainty displayed in standard deviations. The uncertainty in this work mainly comes from the instability of the corona discharge; though the corona current was kept constant at 30 μA for all the experiments, the needle voltage and the electric field distribution might vary for each experiment.

Corona discharge characteristics
The corona current was kept constant at 30 μA using the nineneedle electrode. No pulse current was observed when measuring the current with a matched, 50 Ω coaxial cable. The discharge power at atmospheric pressure was 240 mW for water cathode and 200 mW for mesh cathode. At 100 torr, the discharge power decreased to 93 mW for water cathode and 97 mW for mesh cathode. The corona discharge was also tested with a single-needle electrode, but the d.c. current could not reach 30 μA without causing large current pulses.

H 2 O 2 production in water
The H 2 O 2 production increased linearly with increasing treatment time for tests with both water cathode and mesh cathode. Higher gas flow rate would decrease the concentration of both active neutral species and water vapour in the reactor. Figure 2 illustrates H 2 O 2 production with gas flow rates of 0.002 slpm, 0.02 slpm and 1.0 slpm at atmospheric pressure.
As figure 2 shows, H 2 O 2 production decreased when the gas flow rate increased from 0.002 slpm to 1.0 slpm for both types of cathodes. The rate of decline of H 2 O 2 production with increasing gas flow rate was significant with the mesh cathode, where the positive ions were filtered. At 0.002 slpm and atmospheric pressure, the contribution of active neutral species to H 2 O 2 production was around 50%, dropping to around 25% at 1.0 slpm.
A two-layer mesh cathode was tested in order to verify the ionic wind reduction by the mesh. In the two-layer mesh tests, two meshes were both grounded and placed above the solution surface. The gap between the two meshes was set to 1 mm. The distance from the lower layer to the solution surface was 5.3 mm. The needle tip was 10 mm above the upper mesh. The production of hydrogen peroxide was found to be the same as that with a single-layer mesh cathode. The influence of the mesh pore size was also explored, and there was no significant difference in the production of hydrogen peroxide between mesh cathodes with pore sizes of 1 mm and 10 mm. These results suggested that the mesh did not affect the flow of active neutral species to the solution surface. The results prove that the difference in H 2 O 2 production between the water and mesh cathode was mainly caused by the positive drift ions that were filtered by the mesh.
Increased distance between the mesh and solution was found to reduce the effect of active neutral species on hydrogen peroxide production. H 2 O 2 production decreased by 43.3% when the distance was increased to 33.3 mm; when a squared PTFE disk (with thickness of 1 mm and side length of 60 mm) was situated beneath the mesh covering the whole mesh area, with a distance of 33.3 mm to the water surface, the hydrogen peroxide production was reduced by 90%. The PTFE disk blocked the transportation of active neutral species into the solution.

OH or H 2 O 2 produced?
To investigate the reaction path for hydrogen peroxide formation in the solution, and whether this is via the dimerization of hydroxyl radicals, two types of OH scavengers were used in the experiment: TB (tert − C 4 H 9 OH) and DMSO ((CH 3 ) 2 SO). If the H 2 O 2 in the solution was from dimerization of OH, the addition of TB or DMSO would scavenge OH and reduce the H 2 O 2 production. The reactions between OH and these two scavengers are: Three sample solutions were tested: deionized water, 0.01 mol l −1 TB solution, and 0.01 mol l −1 DMSO solution. Each sample was treated for 30 min at a gas pressure of 300 torr and a gas flow rate of 0.5 slpm. In the tests with TB solution at 760 torr, the corona discharge current could not reach 30 μA without causing pulses. When the gas pressure was reduced to 300 torr, the corona discharge current could maintain at 30 μA.
In the water cathode tests, H 2 O 2 production increased by 64.7% and 40.1% with the addition of TB and DMSO, as figure 3 shows.
In the mesh cathode tests, only active neutral species reached the solution. Hydrogen peroxide production increased by 43.5% with the addition of TB, while the addition of DMSO had minimal effect on H 2 O 2 production.
The OH scavengers did not reduce the H 2 O 2 production, implying that neither positive drift ions nor active neutral species acted to produce OH in the solution. Higher concentrations (0.05 mol l −1 and 0.1 mol l −1 ) of TB and DMSO solution were used, but identical results were obtained. The H 2 O 2 in the solution was possibly transported from the gas phase (H 2 O 2(g) → (H 2 O 2(aq) ).

pH and conductivity
Five millilitres of deionized water were treated for 60 min with a gas flow rate of 0.2 slpm, at 760 torr. The original conductivity of the deionized water was measured to be 0.75 μS cm −1 . After treatment, the conductivity increased to 38.0 μS cm −1 and 41.3 μS cm −1 for the water cathode tests and mesh cathode tests, respectively. The original pH of the deionized water was around 6.40. The pH of the treated water was 4.38 following tests with the water cathode, and 4.60 for the mesh cathode. The conductivity and pH of the treated deionized water for the two types of cathode are very close, indicating that the active neutral species led to the change of conductivity and pH in water. With water cathode, the concentration of NO 2 − and NO 3 − was measured to be 0.035 mmol l −1 and 0.123 mmol l −1 ; with mesh cathode, the concentration of NO 2 − and NO 3 − was 0.043 mmol l −1 and 0.137 mmol l −1 . The pH calculated based upon the concentration of NO 2 − and NO 3 − is 3.80 for the water cathode tests and 3.74 for the mesh cathode tests, which proves that the decrease of pH was mainly caused by the production of nitrite and nitrate. Cadorin et al also found that the pH of water dropped to 3 following the treatment with corona discharge in nitrogen (Cadorin et al 2015). Takahashi et al also reported that NO 2 − and NO 3 − were produced in water after plasma treatment in a N 2 /O 2 gas mixture (Takahashi et al 2016).

Gas pressure effect on H 2 O 2 production
As the gas pressure was reduced, the gap voltage required to maintain a constant 30 μA d.c. current decreased, as shown in figure 4. The gap voltage from needle to cathode was calculated by subtracting the partial voltage across R 2 (which equals to 747 V at 30 μA d.c. current) from the needle voltage measured by the high voltage probe. The gap voltage with mesh cathode is higher than that of water cathode under the same gas pressure. Since the mesh thickness is 0.6 mm, the practical discharge gap could be larger than 10 mm for mesh cathode. The water resistance is approximately 20 kΩ at a conductivity of 1 μS cm −1 . Since the current was 30 μA, the partial voltage across the water was around 600 mV at the beginning of gas discharge. As the gas discharge proceeded, the water conductivity increased and the partial voltage dropped. At 100 torr, the gap voltage for the two cathode types was similar. As the gas pressure was raised, the voltage difference between two cathode types also increased, reaching a maximum of 1.11 kV at the highest tested pressure of 760 torr.
The relationship between H 2 O 2 production and gas pressure was investigated under five different pressures of 100, 200, 300, 500 and 760 torr, as shown in figure 5. Five millilitres of deionized water was treated by positive ionic wind for 30 min, at a gas flow rate of 0.002 slpm.
With water cathode, H 2 O 2 production was 0.148 μmol at both 760 torr and 500 torr. As gas pressure was decreased, the H 2 O 2 production increased, reaching a maximum of 0.250 μmol at 100 torr. H 2 O 2 was produced in water by both the positive drift ions and active neutral species from the corona discharge.
With mesh cathode, only active neutral species reacted with water to produce H 2 O 2 . H 2 O 2 production was much lower than that using the water cathode. The gas pressure did not significantly change H 2 O 2 production, only a slight increase was found between 500 and 100 torr, with a maximum of 0.085 μmol obtained at 100 torr.
The difference in H 2 O 2 production was caused by positive drift ions. The increase of H 2 O 2 production with the deionized water cathode was mainly contributed by positive drift ions as the gas pressure dropped. When the gas pressure dropped from 760 to 100 torr, the average reduced electric field increased by 3 times approximately, resulting in the positive drift ions attaining higher kinetic energy. As figure 6 shows, the energy efficiency of H 2 O 2 production increased as gas pressure decreased.

Effect of drift ions
In the gas phase, the concentration of water vapour is much higher than that of positive nitrogen ions. The positive nitrogen ions would react with water vapour in tens of microseconds (Good et al 1970): Gas-phase water ions H 2 O + g continue to react with water molecules: The produced OH in reaction (4) would be consumed quickly by self-quenching or reacting with nitrogen compounds in the gas. The hydronium ions H 3 O + continue clustering reactions, and form hydrated hydronium cluster ions H + (H 2 O) n . Good demonstrated that almost all of the positive ions are transformed to H + (H 2 O) n after 1 ms of discharge in nitrogen containing traces of water vapour (Good et al 1970). It has been reported that hydroxyl radicals (OH) and active oxygen (O) can be adsorbed on anode surface in traditional electrolysis with specific electrode materials like Boron-doped diamond and Ti/SnO 2 (Marselli et al 2003, Comninellis 1994, which is possibly similar to the effect of the ion anode in this case. These electrode materials are non-reactive and have weak adsorption to OH, leading to low O 2 evolution but high OH production (Marselli et al 2003).   In this research, one mol positive ions can produce only 0.13 mol hydrogen peroxide in water at atmospheric pressure, and the ratio increased to 0.29 when the gas pressure decreased to 100 torr, suggesting that the production of H 2 O 2 was concomitant with O 2 evolution. The hydrogen peroxide production by ions was doubled when TB or DMSO was added as table 1 presents. No free radicals produced by the plasma-water reactions were found able to react with both TB and DMSO to produce H 2 O 2 . One hypothesis is that the adsorbed hydroxyl radicals H + (H 2 O) n (OH) and active oxygen H + (H 2 O) n (O) can synergistically interact with TB and DMSO, doubling the production of hydrogen peroxide.
The mechanisms of H 2 O 2 formation in the plasma-water interactions have been widely investigated. The hydrogen peroxide production was thought mainly from the dimerization of hydroxyl radicals in gas phase and transported to liquid water, as in the gliding arc discharge (Locke and Thagard 2009): The hydroxyl radicals are produced in gas phase by the water dissociation as a result of interactions with electrons, ions and metastables (Fridman 2008, Shainsky et al 2012, Locke and Shih 2011 On the other hand, at the plasma-water interface, the kinetic ions can excite and ionize water molecules to produce OH (Bruggeman et al 2009), and form hydrogen peroxide in the water. Zhao et al investigated the production of hydrogen peroxide in water using pulsed positive discharges in nitrogen with a pin anode and water cathode, and found that the production of hydrogen peroxide decreased significantly with the addition of OH scavenger to water. The decrease of hydrogen peroxide demonstrates that OH is the precursor of hydrogen peroxide, where the interface reactions were initiated by kinetic positive ions (Zhao et al 2016). In contact glow discharge electrolysis, the kinetic ions contributed more OH production than the charge-transfer process, resulting in that the current efficiency of H 2 O 2 production exceeded 100% (Hickling and Ingram 1964). The optical emission spectroscopy also shows there are O, H α and OH produced in gas phase for a pulsed plasma generated in atmospheric pressure air between a pin anode and a tap water cathode (Bruggeman et al 2009).The recombination of electrons with H 2 O + or H 3 O + was thought important in the production of OH.
In this research, the addition of OH scavengers increased the production of hydrogen peroxide, which proves that hydroxyl radicals are not the major reactive products in the reactions between water and positive drift ions.
The drift ions have little kinetic energy, which makes the interface reactions different from the reactions discussed above, and more like an anode process in the electrolysis. The drift ions on water surface break water molecules into the adsorbed (OH) as shown in figure 7, rather than the water excitation and ionization by kinetic ions.

Effects of active neutral species
In this research, with the addition of TB, the production of H 2 O 2 by active neutral species was increased by 50%; with the addition of DMSO, the production of H 2 O 2 was almost unchanged. HO 2 was probably the major active neutral species, since it can react with TB to produce H 2 O 2 , and does not react with DMSO. HO 2 + tert − C 4 H 9 OH → (CH 3 )2C(OH)CH 2 + H 2 O 2 (12) According to reaction (13), two HO 2 can self-react to produce one H 2 O 2 in water. When TB was added in water, one HO 2 can produce one H 2 O 2 . With the same amount of HO 2 , the addition of TB should double the H 2 O 2 production. In considering that the total H 2 O 2 production was increased only by approximately 50% when TB was added, HO 2 self-reaction cannot be the only path for H 2 O 2 production, there should be transportation and dissolution of H 2 O 2 into the solution, formed via OH dimerization in gas phase (Herron and Green 2001): N 2 (A) + H 2 O → H + OH + N 2 (X) N( 2 D) + H 2 O → OH + NH (16)

Conclusions
The research demonstrated that H 2 O 2 is produced in water under exposure to the ionic wind from d.c. corona discharge in nitrogen. The hydrated hydronium drift ions H + (H 2 O) play an important role at the interface reactions with water. These drift ions act as ion anodes, on which the adsorbed hydroxyl radicals (OH) and adsorbed oxygen (O) are formed, capable of initiating advanced oxidation process in water, accompanying the evolution of oxygen and hydrogen gas, and the formation of H 2 O 2 in water. The production of H 2 O 2 in water was linear with treatment time but affected by gas pressure and flow rate, with 1 mol ions producing 0.29 mol H 2 O 2 at 100 torr, equivalent to 47 eV in producing one H 2 O 2 molecule, or an energy yield of 27 g kWh −1 . The overall energy efficiency was still low in considering that the input energy was mostly dissipated in corona discharge.
The active neutral species that led to the production of H 2 O 2 in water is thought to be from the combination of HO 2 as well as the dissolution of gas phase H 2 O 2 . The water pH and conductivity changes observed following the interactions with the ionic wind were mainly caused by the neutral species, which produced nitrite and nitrate ions in water.