Spatio-temporal evolution characteristics and pattern formation of a gas–liquid interfacial AC current argon discharge plasma with a deionized water electrode

The temporal evolution of the argon discharge emission patterns (recorded using a constant-amplitude AC discharge current) over a liquid as a function of the discharge exposure time (t_exp,d) is presented in figure 2. The corresponding root mean square discharge voltage (V_rms,d), root mean square discharge current (I_rms,d), power input (P_in), liquid electrode electrical conductivity (σ_e), and pH value are listed in each panel of figure 2. All images are captured using a camera with a fixed exposure time of t_exp = 100 ms. Immediately after breakdown, the discharge initiated in the gas phase also extends into the liquid (V_rms,d = 3300 V and I_rms,d = 4 mA). Hence in the reactor discharge exists in the gas and liquid phases simultaneously and thus can be called a gas–liquid phase discharge. This gas-phase discharge extension into the liquid is an exceptional phenomenon that has rarely been reported.

When the applied V_rms,d increases, the scale of the discharge increases as well. The discharge column lengthens radially in the gas phase while the liquid discharge region expands (figures 2(b) and (c)). To observe this behavior over time, t_exp,d is increased using a constant-amplitude AC discharge current (I_rms,d = 35 mA). The liquid discharge region shrinks and a cone-like structure appears at the liquid surface (figures 2(c)–(f)). A contraction in the cone-like structure is noted when t_exp,d is extended further (figures 2(f)–(h)). The evolution process described above is accompanied by decreases in P_in, V_rms,d, and the pH, as well as an increase in σ_e (figures 2(c)–(h)).

The gas-phase discharge appears visually like a constricted channel. Initially, the liquid discharge region has a plume-like shape (figure 2(b)). One may think that the plume-like shape of the liquid discharge region results from penetration by flowing gas rather than from propagation of the discharge in the liquid phase. The liquid discharge region is investigated by analyzing the discharge shapes and computer software is utilized to measure the discharge dimensions. (i) First, the hole depth generated in the liquid by the gas flow is measured (approximately 1 mm in figure 2(a)). This is much shorter than the discharge length in the downstream direction in the liquid (approximately 4 mm in figure 2(b)). So, extended discharge after covering approximately 1 mm depth into the liquid surface will no longer remain in the gas phase (inside the liquid). (ii) As the V_rms,d value increases, the in-liquid discharge region spreads in both the downstream and the radial directions (from ~4 to up to ~7 mm and from ~4 to up to ~8 mm, respectively, in the downstream and radial directions). In contrast, conventional plasma jet plumes elongate only in the stream direction in the gas phase. (iii) Visually, the liquid discharge region appears to be highly unstable and to follow random patterns. Thus, it can be concluded that the liquid discharge region is not a gas–phase plume, but rather is a direct liquid discharge. It might be comprised of single or multiple non-stationary streamer channels that spread chaotically through the liquid. The cone-like structure appears later on the liquid surface and seems to be diffuse and homogeneous.

Non-stationary discharge channels in direct liquid discharges have been observed and described in [17, 18]. At lower liquid conductivities, the discharge channels are thinner and longer [17, 32]. A similar phenomenon is observed in the present case; the discharge extension into the liquid is maximized at lower conductivities (DIW initial conductivity σ_e = 3 μS cm⁻¹ in figures 2(b) and (c)). After a minute of operating time, the liquid discharge region starts to shrink as the conductivity increases (from DIW σ_e = 3 μS cm⁻¹ to electrolyte σ_e = 11 μS cm⁻¹), as shown in figures 2(c) and (d). The increase in σ_e is attributed to gas impurities that are transported into the reactor and lead to generation of acidic products in the liquid (see section 3.7). After 2 min of exposure, the in-liquid discharge region vanishes and a cone-like structure appears at the liquid surface. A nearly cone-like discharge structure has been observed in atmospheric air discharges sustained by an AC power supply (5–20 kHz) with a liquid as one of the electrodes [11]. The electrolyte conductivity is an important parameter for discharge structure generation. However, the discharge structure generation mechanism from these previous experiments differs from ours [11]. One dissimilarity is in the experimental procedure. In our case, the cone-like structure appears after a t_exp,d of 2 min. In the previous case, it was observed by raising the discharge power to a certain level shortly after the discharge was formed. In addition, static atmospheric air was used as a plasma-forming gas in the previous experiments. The previous experiments also used liquid electrodes with different electrical conductivities (either DIW or tap water with electrical conductivities of 0.055 μS cm⁻¹ and 500 μS cm⁻¹, respectively) [11]. A similar cone-like pattern was also reported over the DIW (σ_e = 20 μS cm⁻¹) electrode immediately after initiation of an AC-excited (operated at 22 kHz) helium discharge [14].

Gas-phase discharge propagation in the liquid phase is a complex physical phenomenon that can involve various concepts such as strong electric fields, phase instabilities, heating, bubble formation, and molecular dissociation. It has been concluded that the discharge process requires Joule heating to create bubbles in water [33]. At the bubble-water interface, electrohydrodynamic instability can occur when stronger electrical fields are present, resulting in the breakup of bubbles into streamers [33]. In addition, a reduction in the extension of the in-liquid discharge region with t_exp,d can be correlated to an increased liquid electrode σ_e. Dissociation and ionization of water molecules by the applied electric field increase σ_e. In addition, acidic products produced due to the presence of air impurities in the reactor are also involved in increasing σ_e. Thus, the σ_e increases near the discharge channel head and the electric field from that area within the Maxwellian relaxation time becomes banished, and further dissociation does not occur inside the discharge [34]. To develop the discharge channel, additional energy must be provided via the power supply to overcome the delay in restitution of the electric field at the channel head. However, in our case, the discharge power input decreases with t_exp,d (figure 2). When this happens, channel development is interrupted and discharge propagation decreases with t_exp,d (figures 2(c)–(e)). Likewise, the cone-like pattern at the liquid surface shortens with t_exp,d (figures 2(f)–(h)). Therefore, formation of various discharge shapes in the gas and liquid phases is sensitive to variation in liquid properties.

Temporally recorded current–voltage waveforms that correspond to the discharge emission patterns in figures 2(b), (c), (f), and (h) are displayed in figures 3(a)–(d), respectively. The current waveform in figure 3(a) recorded for the discharge (which extends into the liquid) in figure 2(b) shows a single dominant current pulse every half-waveform. It is evident that the current pulse is larger in amplitude and shorter in time (figure 3(a)). The current pulse amplitude continues to decrease as the applied voltage increases and is extinguished completely when the applied voltage is high (figure 3(b)). The current waveforms in figures 3(c) and (d) recorded for the discharges (appeared in a cone-like structures at the liquid surface) in figures 2(f) and (h) are quite different in that the current pulses that appear are shorter in amplitude and longer in time (see figures 3(c) and (d)).

Zoom In Zoom Out Reset image size

As shown in figure 3, the current pulse characteristics from early and later discharge stages are completely different. The disparity in current pulse characteristics can be described as follows. At low power input levels, the discharge column has poor conductivity and the discharge current approaches zero when the discharge voltage changes polarity (figure 3(a)). The observed current spike amplitudes are different at different polarities since one electrode is liquid and the other is metallic (figure 3(a)). When the power input is increased further, a smoother current waveform is observed because more energy is absorbed into the discharge. This implies a higher discharge conductivity (figure 3(b)). Moreover, it is clear that transformation of the discharge into a cone-like structure is accompanied by obvious increase in current pulse durations (figures 3(c) and (d)). Also, the cone-like structure appears during later operating stages at relatively low applied voltages (figure 2). Thus, the critical electric fields during the early and later stages of the discharge are different. The α coefficient (the average number of free electrons produced in a unit distance via collision of an electron with neutral gas atoms and molecules) should be larger for higher electric fields. In this case, the electron avalanche is established rapidly and additional electrons are released, resulting in electron avalanche multiplication. Thus, the charged particle flux is stronger under higher electric fields. Consequently, the generated electric field opposes the applied electric field, and the current pulse duration is shorter when the charged particle flux is stronger. Therefore, the Δt_pulse noted in the current waveforms is short immediately after discharge initiation but a larger Δt_pulse is noted later. The recorded waveforms are unique and can be treated like a signature because of the two distinctive discharge shapes observed in- and at- liquid surfaces over the course of the discharge. The current pulses suggest that the discharges are streamer-like [35]. Further investigation might be conducted using high-speed imaging cameras.

In figure 4, spatially resolved emission spectra are recorded (at constant-amplitude AC current) to compare the discharge chemistry of two distinctive discharge regions; at- and in- liquid surfaces, as shown in the insets of figures 4(a) and (b), respectively). Both spectra are dominated by nitrogen oxide NO· radicals bands for NO-γ (A–X) transitions (occur between 230 and 272 nm), bands of hydroxyl ·OH radicals from OH (A–X) transitions (occur between 306 and 310 nm), and bands of the second positive system of molecular nitrogen for N₂ (C–B) transitions (occur between 330 and 381 nm). Moreover, the hydrogen Balmer alpha (H_α) line (occurs at 656 nm), and atomic oxygen line (occurs at 777 nm) are also detected. As argon is the feeding gas, atomic transitions of Ar I at 696, 706, 738, 752, 763, 772, 794, 801, and 811 nm are visible in both emission spectra in the range of 690–815 nm. In figure 4(a) for the gas–liquid interface region of discharge, molecular NO-γ (A–X), OH (A–X), and N₂ (C–B) bands transitions, and atomic H and O lines transitions are resulted from water vapor and ambient air gas entrainment. Moreover, emission intensities of NO-γ (A–X), OH (A–X), and N₂ (C–B) bands, and also Ar I lines are relatively stronger for the gas–liquid interface region in comparison to the in-liquid discharge region. In fact, there is a lesser exposure to air and feeding gas for the liquid discharge region. On the other hand, a part of the power is utilized in the dissociation of water molecules to propagate the discharge in the liquid phase. The key reaction pathways involved in the formation of various species are discussed in detail.

Zoom In Zoom Out Reset image size

The emission bands such as NO-γ (A–X) and OH (A–X) are main features in the UV region of spectra. The chemical reaction (R1) is recognized to exclusively responsible to the production of NO· radical [23]:

$$\begin{eqnarray}&&{\rm{N}}+\cdot {\rm{OH}}\to \cdot {\rm{H}}+{\rm{NO}}\cdot \end{eqnarray} \tag{ R1 }$$

This reaction boosts up in moist conditions (e.g. at gas–liquid interface), because H₂O molecule Rydberg levels are resonant with ${{\rm{Ar}}}_{{\rm{meta}}}^{* },$ consequently ·OH is efficiently generated [36, 37]. Moreover, using the kinetic modeling, it was shown that, ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ yields a significant amount of N, subsequently stimulates reaction R1 [38]. Therefore, for argon-fed discharge under humid conditions, reaction (R1) is mainly dominant. To a minor extent, formation of NO· is known to consequence through the three-body reaction involving O· and N [38]

$$\begin{eqnarray}&&{\rm{O}}\cdot \,+{\rm{N}}+{\rm{M}}\to {\rm{NO}}\cdot \,+{\rm{M}}\end{eqnarray} \tag{ R2 }$$

and also with the interaction of O· and metastable nitrogen molecule N₂ (A)

$$\begin{eqnarray}&&{\rm{O}}\cdot \,+{{\rm{N}}}_{2}({\rm{A}})\to {\rm{NO}}\cdot \,+{\rm{N}}.\end{eqnarray} \tag{ R3 }$$

The excitation mechanism to NO-γ (A–X) band is through the following reaction of NO (X) with N₂ (A),

$$\begin{eqnarray}&&{\rm{NO}}\left({\rm{X}}\right)+{{\rm{N}}}_{2}\left({\rm{A}}\right)\to {\rm{NO}}\left({\rm{A}}\right)+{{\rm{N}}}_{2}\left({\rm{X}}\right).\end{eqnarray} \tag{ R4 }$$

and then radiative emission to NO-γ (A–X) band

$$\begin{eqnarray}&&{\rm{NO}}\left({\rm{A}}\right)\to {\rm{NO}}\left({\rm{X}}\right)+h\nu .\end{eqnarray} \tag{ R5 }$$

The highly reactive radicals, e.g. OH (A–X) band emissions are recorded at 309 nm. Among the several reactions to the formation of ·OH radical in the humid discharge, one of the probable reactions here is the electron impact dissociation to form ·H and ·OH radicals [39]:

$$\begin{eqnarray}&&{{\rm{e}}}^{-}+{{\rm{H}}}_{2}{\rm{O}}\to \cdot {\rm{OH}}+\cdot {\rm{H}}.\end{eqnarray} \tag{ R6 }$$

The ionization of water molecules requires relatively higher electron energies (12.6 eV), so electron impact dissociation (6.4 eV) seems feasible reaction pathway for ·OH production. Dissociative recombination of H₂O⁺ was hypothesized as a possible route for ·OH production [40]:

$$\begin{eqnarray}&&{{\rm{e}}}^{-}+{{\rm{H}}}_{2}{{\rm{O}}}^{+}\to \cdot {\rm{OH}}+\cdot {\rm{H}}.\end{eqnarray} \tag{ R7 }$$

In this reaction, H₂O⁺ is less-likely the outcome of electron impact dissociation. The ionization potentials of H₂O and N₂ are 12.6 and 15.6 eV, respectively. In the presence of substantial water molecules in plasma, relatively lower ionization of H₂O is more favorable to originate H₂O⁺ ion through the Penning ionization process

$$\begin{eqnarray}&&{{\rm{Ar}}}_{{\rm{meta}}}^{* }+{{\rm{H}}}_{2}{\rm{O}}\to {\rm{Ar}}+{{\rm{H}}}_{2}{{\rm{O}}}^{+}+{{\rm{e}}}^{-}.\end{eqnarray} \tag{ R8 }$$

Another most probable pathway to produce ·OH radical is the direct dissociative excitation of H₂O through the ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ [36, 37]

$$\begin{eqnarray}&&{{\rm{Ar}}}_{{\rm{meta}}}^{* }\left[{}^{3}{\rm{P}}_{0}(1{{\rm{s}}}_{3})\right]+{{\rm{H}}}_{2}{\rm{O}}\to {\rm{Ar}}+{\rm{OH}}\left({\rm{A}}\right)+{\rm{H}},\end{eqnarray} \tag{ R9 }$$

$$\begin{eqnarray}&&{{\rm{Ar}}}_{{\rm{meta}}}^{* }\left[{}^{3}{\rm{P}}_{2}(1{{\rm{s}}}_{5})\right]+{{\rm{H}}}_{2}{\rm{O}}\to {\rm{Ar}}+{\rm{OH}}\left({\rm{A}}\right)+{\rm{H}}.\end{eqnarray} \tag{ R10 }$$

Reactions (R9) and (R10) are considered essential because ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ levels are resonant with Rydberg levels of water molecule, and subsequently ·OH radical production is boosted in the existence of humidity. Although, all the above-mentioned reaction pathways (R6)–(R10) are important for the ·OH radical emission in the discharge region at the gas–liquid interface. Nevertheless, for liquid discharge region, lesser exposure to air and feeding gas suggests that electron impact reaction (R6) is principally more critical to produce ·OH radical. Aside from that, the nitrogen plasma emission band of the radiative state N₂ (C–B) can be resulted from electron impact excitation [41]

$$\begin{eqnarray}&&{{\rm{e}}}^{-}+{{\rm{N}}}_{2}\left({\rm{X}}\right)\to {{\rm{N}}}_{2}\left({\rm{C}}\right)+{{\rm{e}}}^{-},\end{eqnarray} \tag{ R11 }$$

and also from the quenching of N₂ (A) molecules through Pooling reaction [42]

$$\begin{eqnarray}&&{{\rm{N}}}_{2}\left({\rm{A}}\right)+{{\rm{N}}}_{2}\left({\rm{A}}\right)\to {{\rm{N}}}_{2}\left({\rm{X}}\right)+{{\rm{N}}}_{2}\left({\rm{C}}\right),\end{eqnarray} \tag{ R12 }$$

where important metastable N₂ (A) can be the outcome of ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ through following reaction [38]:

$$\begin{eqnarray}&&{{\rm{Ar}}}_{{\rm{meta}}}^{* }+{{\rm{N}}}_{2}\left({\rm{X}}\right)\to {\rm{Ar}}+{{\rm{N}}}_{2}\left({\rm{A}}\right),\end{eqnarray} \tag{ R13 }$$

and ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ also causes a reaction with N₂ (X) that originates into the radiative state of N₂ (C–B) [43]

$$\begin{eqnarray}&&{{\rm{Ar}}}_{{\rm{meta}}}^{* }+{{\rm{N}}}_{2}\left({\rm{X}}\right)\to {\rm{Ar}}+{{\rm{N}}}_{{\rm{2}}}\left({\rm{C}}\right).\end{eqnarray} \tag{ R14 }$$

The ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ states energies [${}^{3}{\rm{P}}_{2}$ (11.5 eV) and ${}^{3}{\rm{P}}_{0}$ (11.7 eV)] are higher than threshold energy required to excite the N₂ molecule (11.1 eV). Thus, the inclusion of nitrogen to argon leads to an increased emission intensity of N₂ (C–B) through the energy transfer between ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ and N₂ (X) (reactions R12–R14). Note here, in the emission spectra presented in figure 4, nitrogen first negative system ${{\rm{N}}}_{2}^{+}$ (B–X), which is by far most common in atmospheric plasmas is not visible at all. Though, it is visible for other locations of discharge as shown in the spatial measurements in section 3.5. The ${{\rm{N}}}_{2}^{+}$ (B–X) band emission requires higher electron energies over its 18.8 eV threshold for the N₂ (X) molecule in the ground state. In excessive water vapor environment, the strong emissions of OH (A–X) and N₂ (C–B) are likely to be expected through ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ (reactions (R7)–(R10) and (R12)–(R14)) compared to ${{\rm{N}}}_{2}^{+}$ (B–X) emission. It also indicates that an increase in water vapor concentration favors the formation of ·OH radical (see section 3.5). It is commonly accepted that atomic O· is originated from the dissociation of H₂O in a direct-in-liquid discharge:

$$\begin{eqnarray}&&{{\rm{e}}}^{-}+{{\rm{H}}}_{2}{\rm{O}}\to {{\rm{H}}}_{2}+{\rm{O}}\cdot \,+{{\rm{e}}}^{-}.\end{eqnarray} \tag{ R15 }$$

However, exposition of discharge to ambient air gas suggests that the occurrence of O· is not only the result of H₂O dissociation but also of O₂ [44, 45]:

$$\begin{eqnarray}&&{{\rm{e}}}^{-}+{{\rm{O}}}_{2}\to 2{\rm{O}}\cdot \,+{{\rm{e}}}^{-}.\end{eqnarray} \tag{ R16 }$$

Besides, for an argon assisted discharge, involvement of ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ to produce O· cannot be ignored through the following reaction [38]:

$$\begin{eqnarray}&&{{\rm{Ar}}}_{{\rm{meta}}}^{* }+{{\rm{O}}}_{2}\to {\rm{Ar}}+2{\rm{O}}\cdot \end{eqnarray} \tag{ R17 }$$

From the previous discussion on the various chemical reaction pathways, it is evident that ambient air and water vapor entrainment into the discharge regions lead to an active underpinning of plasma chemistry, and ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ states are participating the most to shape the chemistry suitable for the production of various TRS. Meanwhile, higher emission intensities of OH (A–X), NO-γ (A–X), and N₂ (C–B) transition bands at the gas–liquid interface proposed that discharge chemistry is stronger here than the in-liquid discharge region.

The temporal development patterns of the TRS and ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ emission intensities at the gas–liquid interface are shown in figure 5. This discharge region is selected because of its importance in TRS formation, as indicated in section 3.5. Emission spectra are collected in one minute intervals. Molecular transition bands such as those of NO-γ [(A–X), (0,1)] at 236 nm, N₂ [(C–B), (0,0)] at 337 nm, and OH (A–X) at 309 nm, and atomic lines such as ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ (${}^{3}{\rm{P}}_{2}$) at 811 nm and O at 777 nm are considered. An obvious drop in emission intensities is seen at a t_exp,d of 1–2 min. At the same time, a substantial decrease in the power input from 191 to 73 W is recorded (figure 2). This suggests a decrease in the density of highly energized electrons. These electrons are not only thought to be involved in excitation, dissociation, and ionization of N₂, O₂, and H₂O via impact reactions but are also responsible for exciting argon gas atoms to metastable states. As the discharge time passes, all of the spectral lines exhibit steady trends that correspond well with the small discharge power input decreases that occurred later (figure 2). In an argon-fed discharge, where TRS can also be formed in addition to the impact reactions of N₂, O₂ or H₂O (reactions (R11), (R15), and (R16)) through Penning ionization or excitation and dissociative excitation of ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ (reactions (R7)–(R10) and (R12)–(R14)). This also means that decrease in TRS emission intensities for reasons other than power input changes can be the result of relatively low ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ densities.

Zoom In Zoom Out Reset image size

For a deep understanding of TRS in various discharge locations, we can measure the spatial profiles of TRS emission intensities via OES to identify and monitor TRS in real time. Thus, the spatial distributions of emission intensities are recorded for the gas-phase discharge between the cylindrical electrode tip and the DIW surface using the optical arrangement in figure 1(b). The results are presented in figure 6. Emission data is collected in 1.0 mm intervals from the cylindrical electrode tip to the liquid surface. Experiments are performed just after the discharge exposure time of 10 min when the discharge voltage became stable. Various spectral line profiles including NO-γ [(A–X), (0,1)] at 236 nm, N₂ [(C–B), (0,0)] at 337 nm, ${{\rm{N}}}_{2}^{+}$ [(B–X) (0,0)] at 390 nm, OH (A–X) at 309 nm, ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ (${}^{3}{\rm{P}}_{2}$) at 811 nm, and O at 777 nm are considered.

Zoom In Zoom Out Reset image size

Starting from the cylindrical electrode tip, the emission intensities are almost steady because the quartz tube provides a nearly inert discharge atmosphere. Outside the tube, considerable variation in spectral intensities is recorded in humid ambient air. The discharge imports various molecular impurities (e.g. N₂, O₂, and water molecules) that eventually trigger various chemical reactions. The spatial profiles demonstrate that the NO-γ (A–X), N₂ (C–B), and OH (A–X) intensities increase, while the ${{\rm{N}}}_{2}^{+}$ (B–X), ${{\rm{Ar}}}_{{\rm{meta}}}^{* },$ and O emission intensities decrease as one moves from the tube exit to the liquid surface.

The reactions responsible for increasing OH (A–X) emissions are (R7)–(R10), in which ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ atoms play a vital role in ·OH formation. The Penning ionization reaction (R8) fuels (R7) and direct dissociative excitation of H₂O via reactions (R9) and (R10). Moreover, ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ is involved in the N₂ (C–B) transition. This occurs directly via reaction R14 and indirectly by creating metastable N₂ (A) to stimulate the famous Pooling reaction (reactions (R12) and (R13). OH (A–X) and N₂ (C–B) emissions are most prevalent near the liquid surface. At high water vapor concentrations, ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ is most likely to produce OH (A–X) and N₂ (C–B) transitions via reactions (R7)–(R10) and (R12)–(R14) instead of the ${{\rm{N}}}_{2}^{+}$ (B–X) transition, as discussed earlier in section 3.3. Also, the ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ line intensity decreases while the OH (A–X) and N₂ (C–B) intensities are maximized. This strongly suggests the involvement of chemical reactions that include species with lower energy thresholds than ${{\rm{Ar}}}_{{\rm{meta}}}^{* }.$ Therefore, in a humid argon discharge, dissociative excitation and Penning ionization and excitation are assumed to be likely reasons for the appearance of OH (A–X) and N₂ (C–B) and the disappearance of ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ . The NO· is produced via reactions (R1)–(R3), as discussed in section 3.3. Under humid conditions, the increasing NO· trend can be associated with enhanced ·OH production, which stimulates reaction (R1). Also, ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ commonly leads to N₂ (A) states via reaction (R13) when excessive water molecule impurities are present. This is assumed from the strong N₂ (C–B) emission at the gas–liquid interface. This implies the formation of NO· radicals via reaction (R3). On the other hand, enhanced NO· radical production also reduces the O· atomic line via reactions (R2) and (R3).

Recently, a novel approach was used that combined experimental and numerical approaches to studying the reaction kinetics of non-thermal atmospheric argon plasma jets [38]. Experimental and modeling results suggested that ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ (${}^{3}{\rm{P}}_{2}$) was the major species and that reactive oxygen and nitrogen species originated primarily from ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ reactions with N₂, O₂, and H₂O [38]. There might be other chemical processes that lead to TRS production in the present case. Some of these processes are described in section 3.3. However, qualitative analyses of optical emission spectra (figure 4), the temporal evolutions of various species (figure 5), and spatial profiles (figure 6) suggest that ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ reactions might contribute the most to TRS formation.

To estimate the T_g, OH (A–X) band transition is of particular interest in the case of discharges in- and at- liquid surfaces. These are typically containing excessive water vapor concentrations. It was shown that the rotational population distribution of higher rotational numbers is significantly overpopulated and deviated from Boltzmann distribution in a higher water vapor content [13, 19, 43]. Also, the quenching effect of OH (A–X) through water diminishes the OH (A) effective life time, resulting in a deviation from the Boltzmann rotational population distribution. To avoid from an overestimation of T_g while using the OH (A–X) band, a two-temperature fitting technique was employed to the rotational population distribution. Where a significant overpopulation was witnessed for the higher rotational numbers (starting from J = 13) [19]. The slope of initial fit involving lower rotational numbers was taken as the T_r of OH (A–X). Alternatively, in the present work, population of lower rotational states with single fitting is considered for determination of the T_r. Since, it is believed that the phenomenon of overpopulation is not strong for these rotational states.

For the first electronic state, the spectrum of OH (A–X) mainly contains three branches [Q₁, R₂, and P₁]. To obtain the T_r, Q₁ branch of the OH (A–X) band ro-vibrational system $\left({\rm{A}}{}^{2}{\rm{\Sigma }}^{+},\upsilon ^{\prime} =0\to {\rm{X}}{}^{2}{\rm{\Pi }}_{3/2},\upsilon ^{\prime\prime} =0\right)$ is selected because it is well resolved and situated in an uncongested region of the OH (A–X) spectrum. Lower rotational states of Q₁ branch (involving $(N^{\prime} =1,J^{\prime} =3/2)$ to $(N^{\prime} =6,J^{\prime} =13/2),$ note that state $(N^{\prime} =3,J^{\prime} =7/2)$ is omitted from the calculation) are adopted to minimize the overpopulation's consequences as mentioned earlier. The Boltzmann plot method is employed to extract the T_r from the rotational structure of Q₁ branch. The spectral line emission intensity as a result of rotational transition is given by [46]

$$\begin{eqnarray}&&{I}_{J^{\prime} J^{\prime\prime} }=\displaystyle \frac{{C}_{{\rm{em}}}{\sigma }_{J^{\prime} J^{\prime\prime} }^{4}}{{Q}_{{\rm{r}}}}(2J^{\prime} +1)\exp \left[\displaystyle \frac{-{B}_{\upsilon ^{\prime} }J^{\prime} (J^{\prime} +1)hc}{k{T}_{{\rm{rot}}}}\right],\end{eqnarray} \tag{ 1 }$$

where factor ${B}_{\upsilon ^{\prime} }J^{\prime} (J^{\prime} +1)hc$ denotes the rotational energy $F^{\prime} (J^{\prime} ),$ k is the Boltzmann constant, h is the Plank's constant, Q_r is the rotational partition function, ${\sigma }_{J^{\prime} J^{\prime\prime} }$ refers to the wave number for the transition $J^{\prime} \to J^{\prime\prime} ,$ and ${C}_{{\rm{em}}}$ is the emission coefficient. For the Boltzmann plot, logarithm is applied on both sides of equation (1)

$$\begin{eqnarray}\begin{array}{lll}\mathrm{ln}({{I}}_{J^{\prime} J^{\prime\prime} })-\,\mathrm{ln}(2J^{\prime} +1) & = & C-\displaystyle \frac{{B}_{\upsilon ^{\prime} }J^{\prime} (J^{\prime} +1)hc}{k{T}_{{\rm{r}}}}\\ \,{\rm{with}}\,C & = & \mathrm{ln}\left(\displaystyle \frac{{C}_{{\rm{em}}}{\sigma }_{J^{\prime} J^{\prime\prime} }^{4}}{{Q}_{{\rm{r}}}}\right),\end{array}\end{eqnarray} \tag{ 2 }$$

and left-hand side of the equation (2) is plotted verses $J^{\prime} (J^{\prime} +1),$ where slope of the straight line represents the T_r. Boltzmann plot offers several advantages, such as it improves the accuracy by taking number of lines into consideration, and minimizes possible errors due to involvement of lines weaker in intensity or overlapping. It also facilitates to authenticate whether rotational equilibrium occurs. Though to apply this method, some precautions are requisite such as utilization of well resolved lines free from self-absorption and overlapping. Self-absorption is one of the serious errors, which generates a dip at middle of the plot by the line with strongest emission intensity, and thus results in altering the slope, and leading to an apparent temperature. Therefore, involved lines overlapping with almost equivalent intensity need to be eliminated from the calculation to get rid of from the error. Figure 7(a) shows the Q₁ branch lines included in the calculation and their corresponding Boltzmann plot. The spectrum in figure 7(a) is recorded under the conditions as in figure 6, and it corresponds to the discharge region near the cylindrical electrode tip. It is noted that Q₁(3) line is excluded from the Boltzmann plot because of overlapping with an equivalent emission intensity line Q₂₁(3) on the basis of description as mentioned earlier. It is evident from figure 7(a) that selected lines Q₁(1), Q₁(2), Q₁(4), Q₁(5), and Q₁(6) of the Q₁ branch of OH (A–X) band are in good agreement with linear fitting. So the obtained rotational temperature T_r = 1035 K can be equal to T_g.

Zoom In Zoom Out Reset image size

For comparison with the Boltzmann plot method, the T_r of OH (A–X) is also determined via LIFBASE [47]. This commonly used technique involves fitting an experimental spectrum to a simulated one. The spectrum with the best fit indicates the value of T_r, which is equal to T_g when the rotational states follow the Boltzmann distribution. As an example, a LIFBASE fit to the spectrum in the inset of figure 7(a) is shown in figure 7(b). Good agreement between the experimental and synthetic spectra demonstrates that the rotational states are equilibrated and the resulting rotational temperature T_r = 1050 K is equal to T_g. One can conclude from figure 7 that the T_r values determined using the Boltzmann plot and LIFBASE methods are in good agreement near the cylindrical electrode tip.

The atmosphere in the reactor is humid during the discharge. Thus, one can expect water vapor concentrations to be higher near the liquid surface than at locations far from the liquid surface. As mentioned earlier, the OH (A–X) higher rotational state population departs from the Boltzmann distribution in a humid environment. Therefore, the Boltzmann plot and LIFBASE methods estimated T_r are compared for different discharge locations to approximate the T_g. In that aspect, spatially resolved profiles of T_r of OH (A–X) between the cylindrical electrode tip and liquid surface are deduced by considering these methods (figure 8). Spatial measurements are performed under the same operating parameters as in figure 6. At low water vapor concentrations, there is good correspondence among the T_r profiles inside the quartz tube. This indicates that the rotational states follow the Boltzmann distribution and one can confidently state that T_r equals T_g. As one moves along the discharge axis towards the liquid surface, growth in the T_r profiles is observed via both methods. The rotational temperature estimated using the LIFBASE method increases faster than that determined using the Boltzmann plot method. The maximum T_r values are extracted near the liquid surface, although the T_r calculated using the LIFBASE method (T_r = 3100 K) is much higher than that from the Boltzmann plot method (1510 K). It means that the phenomenon of overpopulation in large water vapor concentrations is not strong for the Boltzmann plot $(N^{\prime} \leqslant 6,J^{\prime} \leqslant 13/2).$ For example, at a point 12.0 mm from the cylindrical electrode tip, the spectral databases calculated using a Boltzmann distribution produce T_r values that vary from 1035 to 1125 K instead of to 1050–1565 K for the LIFBASE method. The marginally higher T_r produced using the Boltzmann plot can still be used as T_g, but the corresponding LIFBASE-estimated T_r is quite high. Nevertheless, the T_r value determined using a Boltzmann plot reaches 1510 K near the liquid surface. Thus, T_r is no longer representative of T_g. This shows that near the liquid surface, selected lines from the Q₁ branch are not aligned upon linear fitting. A typical example of Boltzmann plot fitting to the spectrum (recorded near the liquid surface under the same conditions as in figure 6) in the inset of figure 9(a) indicates the absence of rotational equilibrium. At high water vapor concentrations, considerable overestimation of T_r using the LIFBASE method suggests that the water vapor concentration controls the population distributions of rotational states with high numbers. It also implies that LIFBASE fitting to the OH (A–X) spectrum does not work well when the water vapor content is high. A typical example of LIFBASE fitting to the spectrum in the inset of figure 9(a) is presented in figure 9(b). A poor fit between the experimental and simulated spectra indicates that the estimated T_r does not equal T_g. It is evident from the results that utilization of lower rotational states limits overpopulation, although the hypothesis that T_r is equivalent to T_g does not remain valid near the liquid surface.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The ICs of LRS during the t_exp,d are shown in figure 10. The evolution pattern indicates that ICs of LRS increase in the first 10 min of t_exp,d. Further prolonging of t_exp,d, yields in decreasing ICs of H₂O₂ and ${{\rm{NO}}}_{{\rm{2}}}^{-}.$ In contrast, at the same time, the IC of ${{\rm{NO}}}_{3}^{-}$ increases continuously with t_exp,d. After the dissolution of TRS (e.g. ·OH, ·H, NO·, NO₂·, and O·) in the DIW, the ICs of LRS (e.g. H₂O₂, ${{\rm{NO}}}_{{\rm{2}}}^{-},$ and ${{\rm{NO}}}_{3}^{-}$) can be expected to increase through the following reactions (R18)–(R23) [15, 48]:

$$\begin{eqnarray}&&2{\rm{NO}}\cdot \,+{{\rm{O}}}_{2}\to 2{{\rm{NO}}}_{2}\cdot ,\end{eqnarray} \tag{ R18 }$$

$$\begin{eqnarray}&&{{\rm{NO}}}_{2}\cdot \,+{\rm{NO}}\cdot \,+{{\rm{H}}}_{2}{\rm{O}}\to 2{{\rm{NO}}}_{2}^{-}+2{{\rm{H}}}^{+},\end{eqnarray} \tag{ R19 }$$

$$\begin{eqnarray}&&{{\rm{H}}}_{2}{\rm{O}}+2{{\rm{NO}}}_{2}\cdot \,\to 2{{\rm{H}}}^{+}+{{\rm{NO}}}_{2}^{-}+{{\rm{NO}}}_{3}^{-},\end{eqnarray} \tag{ R20 }$$

$$\begin{eqnarray}&&4{\rm{NO}}\cdot \,+2{{\rm{H}}}_{2}{\rm{O}}+{{\rm{O}}}_{2}\to 4{{\rm{NO}}}_{2}^{-}+4{{\rm{H}}}^{+},\end{eqnarray} \tag{ R21 }$$

$$\begin{eqnarray}&&4{{\rm{NO}}}_{2}\cdot \,+2{{\rm{H}}}_{2}{\rm{O}}+{{\rm{O}}}_{2}\to 4{{\rm{NO}}}_{3}^{-}+4{{\rm{H}}}^{+},\end{eqnarray} \tag{ R22 }$$

$$\begin{eqnarray}&&\cdot {\rm{OH}}+\cdot {\rm{OH}}\to {{\rm{H}}}_{2}{{\rm{O}}}_{2}.\end{eqnarray} \tag{ R23 }$$

The produced LRS are acidic in nature and bring a drop in pH value (from neutral to acidic) and an increase in σ_e (from DIW to electrolyte) of the plasma-treated DIW (see figures 2 and 10). This changing nature of liquid promotes several other chemical reactions in the liquid phase. For example, under acidic conditions (pH < 3.5), reaction R24 takes place [49]:

$$\begin{eqnarray}&&2{{\rm{NO}}}_{2}^{-}+{{\rm{H}}}_{2}{\rm{O}}+{{\rm{H}}}^{+}\to {\rm{NO}}\cdot \,+{{\rm{NO}}}_{3}^{-}+{{\rm{H}}}_{3}{{\rm{O}}}^{+}.\end{eqnarray} \tag{ R24 }$$

It is one of the main reactions responsible for disproportionation of ${{\rm{NO}}}_{{\rm{2}}}^{-}.$ The formation of various LRS as a result of gas phase discharge was studied on dependence with various gas mixtures and pH of the treated solution [15]. According to the post-discharge analysis, the treated solution with fixed pH 6.9 using a buffer has shown an increased tendency of the H₂O₂, ${{\rm{NO}}}_{{\rm{2}}}^{-},$ and ${{\rm{NO}}}_{3}^{-}$ concentrations [15]. In their case, disproportionation reactions are less-likely possible. In our work, DIW with pH 7.5 is used as a treated liquid. Samples are taken from a liquid continuously treated by the gas-phase discharge. Prolonging the t_exp,d over a period of 10 min when the liquid pH value drops to acidic (pH < 3.5), a significant drop in IC of ${{\rm{NO}}}_{{\rm{2}}}^{-}$ follows the further treatment time (figure 10). The drop in concentration of ${{\rm{NO}}}_{{\rm{2}}}^{-}$ corresponds well with the reason that reaction (R24) is promoted under acidic conditions. Besides, the decomposition of H₂O₂ may proceed through reaction (R25) [15]:

$$\begin{eqnarray}&&{{\rm{H}}}_{2}{{\rm{O}}}_{2}+{{\rm{NO}}}_{2}^{-}+{{\rm{H}}}^{+}\to {\rm{O}}={\rm{NOOH}}+{{\rm{H}}}_{2}{\rm{O}}.\end{eqnarray} \tag{ R25 }$$

Reaction (R25) cannot justify the decrease in IC of H₂O₂ later than 15 min of the t_exp,d, as the corresponding IC of ${{\rm{NO}}}_{{\rm{2}}}^{-}$ than that of H₂O₂ is very low. Another reason for H₂O₂ disintegration can be the interaction with NO· through reaction (R26) [50]:

$$\begin{eqnarray}&&3{{\rm{H}}}_{2}{{\rm{O}}}_{2}+2{\rm{NO}}\cdot \,\to 2{{\rm{NO}}}_{3}^{-}+2{{\rm{H}}}_{2}{\rm{O}}+2{{\rm{H}}}^{+}.\end{eqnarray} \tag{ R26 }$$

A well-known process that can play a critical role for the destruction of H₂O₂ is the photolysis

$$\begin{eqnarray}&&{{\rm{H}}}_{2}{{\rm{O}}}_{2}+hv\to 2\,\cdot {\rm{OH}}\end{eqnarray} \tag{ R27 }$$

by UV radiations produced by the discharge [51]. Photolysis process results in the dissociation of H₂O₂ into ·OH. The produced ·OH radical from this process further decomposes H₂O₂ radical via reactions (R28) and (R29) [51]:

$$\begin{eqnarray}&&\cdot {\rm{OH}}+{{\rm{H}}}_{2}{{\rm{O}}}_{2}\to {{\rm{H}}}_{2}{\rm{O}}+{{\rm{HO}}}_{2}\cdot \,,\end{eqnarray} \tag{ R28 }$$

$$\begin{eqnarray}&&{\rm{H}}\,{{\rm{O}}}_{2}\cdot \,+{{\rm{H}}}_{2}{{\rm{O}}}_{2}\to \cdot {\rm{OH}}+{{\rm{H}}}_{2}{\rm{O}}+{{\rm{O}}}_{2}.\end{eqnarray} \tag{ R29 }$$

It is obvious from figure 10 that σ_e substantially increases with longer t_exp,d. However, with increasing σ_e, the production of H₂O₂ by the discharge decreased because of increasing photolysis of H₂O₂ [51]. Thus, in present case, decreasing IC of H₂O₂ with t_exp,d can be associated to the increasing photolysis of H₂O₂.

Zoom In Zoom Out Reset image size

Present results are similar to our previous work on an AC-operated helium discharge over the DIW (20 μS cm⁻¹) electrode, where ICs of H₂O₂ and ${{\rm{NO}}}_{{\rm{2}}}^{-}$ decreased with discharge time along with increased IC of ${{\rm{NO}}}_{3}^{-}.$ It signifies that DIW can also be activated chemically at reduced cost by using inexpensive argon gas. ICs of LRS are significantly dependent on the properties of treated liquid, as the disproportionation and photolysis processes are highly influenced by pH value and σ_e. It is unlike the gas-phase discharge chemistry that is mainly fueled by the electron impact reactions and ${{\rm{Ar}}}_{{\rm{meta}}}^{* }$ to generate various TRS at the liquid surface (see sections 3.3 and 3.5).