Spatio-temporal profile of atomic oxygen in a 1 kHz repetition atmospheric-pressure plasma jet in He–O2–H2O mixture

Atomic oxygen (O) is one of the essential reactive species in plasma oxidation processes. We investigated the behavior of atomic oxygen in a 1 kHz-repetition pulsed plasma jet in atmospheric-pressure He/O2/H2O mixture. By two-photon absorption laser-induced fluorescence, the spatio-temporal profiles of O density were measured under various conditions. In the dry ([H2O] ⩽ 100 ppm) condition, the rate of O production did not depend on the [O2] fraction in the range of [O2] = 275–8600 ppm. The analysis of the O-production rate indicates that the atomic oxygen in this plasma jet arises from electron-impact dissociation and quenching of O(1 D), similar to the O-production mechanism in radio-frequency plasma jet. The dependence of O-production in each discharge pulse (Δ[O]) on the discharge energy Ed and [O2] in the plasma region at dry condition is formulated as [ΔO](cm−3)/Ed(mJ)= 1.3×1015×{1−exp(−1.85×10−17[O2](cm-3))} . The decay rate of atomic oxygen was not explained by self-recombination or ozone-generation reactions; it was consistent with the reaction rate of O + OH → O2 + H at [OH] = 2×1013  cm−3. This result suggests that the small amount of [OH] with 1013 cm−3 density is more responsible for O behavior than [O2] with large fraction of 1015 cm−3. We conducted a chemical reaction simulation considering the measured results of [O] and [OH] production, resulting in good agreement with the spatial distribution of [O]. Chemical reaction analysis revealed that the cyclic reproduction of OH via chain reaction with O and O2 is important, therefore a small amount of OH catalytically consumes atomic oxygen with two-order higher density.

Helium APPJ is generally produced by applying a high voltage between two electrodes wrapped on a dielectric tube, and the plasma flows out of the tube into the ambient air. Hereafter, we refer to the area between the two wrapped electrodes as the 'plasma region', and the luminous region outside the electrodes as the 'plasma plume'. Several studies have reported on the measurement of reactive species in He-APPJ such as He metastable (He * ) [15,16], OH [17,18], O [19,20], O 2 (a 1 ∆) [21,22], and O 3 [23], as these species play major roles in the chemical treatment by the plasma jet. In the previous studies, the reactive species were mostly measured in the plasma plume, and not in the plasma region. This is because the behavior of the reactive species in the plume is more important in evaluating the flux on the target. However, some studies have shown that the reactive species in the upstream plasma region can affect the density of the reactive species in the plume [22,24]. Therefore, the behavior of reactive species in the plasma region is also worth clarifying. Ono and Tokuhiro investigated the behavior of OH from the upstream plasma to the downstream plume in repetitive pulsed He-APPJ [25], revealing the production and loss processes of OH and the dependence of the OH density on the discharge energy and repetition frequency.
Atomic oxygen (O) is one of the most important radicals in oxidation processes using APPJ. Previous studies have reported on the behavior of O in RF-plasma jets [26][27][28], whereas the behavior of atomic oxygen in kHz-repetition pulsed He-APPJ has not been clarified. In this study, we investigated the production and loss mechanism, spatio-temporal distribution for the entire region of the plasma and the plume, and the dependence on discharge energy and O 2 fraction of atomic oxygen in kHz-repetition pulsed APPJ in He/O 2 /H 2 O mixture.

Two-photon absorption laser-induced fluorescence (TALIF) measurement of atomic oxygen
Ground state atomic oxygen O(2p 4 3 P) was measured by TALIF technique [29][30][31][32][33]. Energy levels related to TALIF of O(2p 4 3 P) are represented in figure 1. A 225.6 nm laser excites the ground state O to O(2p 3 3p 3 P) state via two-photon absorption, and an 844.6 nm fluorescence corresponding to 2p 3 3p 3 P → 2p 3 3s 3 S transition is observed as a signal. The TALIF signal S F is integrated over time, volume, and wavelength, and expressed in terms of the O(2p 4 3 P) density n O and laser intensity I as below: where A F , Γ, and σ (2) are the emission coefficient of the observed fluorescence, total deexcitation rate of the excited state, and two-photon absorption cross section, respectively. Hereafter, the word 'atomic oxygen' refers to the ground state O(2p 4 3 P).
The absolute density of atomic oxygen was calibrated by TALIF of xenon (Xe) [34][35][36][37][38]. A standard gas of Xe(1%)/N 2 was used as the reference, and a 224.2 nm laser excited the ground state Xe to 6p ′ [3/2] 2 state, after which an 834.7 nm fluorescence was observed. The absolute value of n O is estimated by Here, a X , E L−X , and C X in equation (2) represent the branching ratio, laser energy per pulse, and fluorescence collection efficiency including the optics, bandpass filters, and detector, respectively, related with species X. The branching ratio a is expressed in terms of the quenching coefficient k i of collider species M i as where A e is the total radiative decay rate of the excited state. The ratio A F /A e is 1 for O (2p 3 3p 3 P) [34,37] and 0.733 for Xe (6p ′ [3/2] 2 ) [37,39]. The parameter f J is the fine structure distribution factor of the ground state O (2p 4 3 P J , J = 0, 1, 2), where U J represents the internal energy of each state. In this study, J = 2 and gas temperature T g = 300 K were used. The  [34] coefficient values in equation (2) used in this study are summarized in table 1. The Xe-TALIF signal dependences on the reference gas pressure and the laser energy were investigated before the calibration, and the signal was saturated at the reference gas pressure of approximately 30 kPa and the laser energy of 0.2 mJ. The calibration was conducted under the condition where S F−Xe was proportional to the partial pressure of Xe and E L 2 . Therefore, the three-body quenching of excited Xe [37] and the Xe-TALIF saturation against laser energy were both negligible.

Experimental settings
The experimental apparatus is presented in figure 2. A mixture of O 2 /H 2 O/He was supplied to a synthesized quartz tube with inner and outer diameters of 4 and 6 mm, and pulsed plasma jet was generated by applying a 1 kHz repetitive high voltage pulse between the two electrodes wrapped on the quartz tube. The applied voltage had a rectangular form with a duration of approximately 800 ns and a peak height of V(0-10 variable) (kV). The applied voltage and current were measured by a high-voltage probe (P6015A, Tektronix) and a current transformer (CT: model 2877, Pearson), respectively. The two electrodes were 10 mm wide and placed 20 mm apart. The total gas flow rate was controlled to be 1 SLM by mass flow controllers, and the volume fraction of O 2 and H 2 O were measured by a zirconia oxygen analyzer (OX400, Yokogawa) and a dew point transmitter (TK-100, Tekhne), respectively. Humidity was not completely eliminated owing to the small amount of gas transmission through the tube and the attached water molecules inside the gas tube. To control the gas composition accurately, the plasma jet plume was shielded from the ambient air by the quartz tube. A grounded electrode was wrapped at the end of the quartz tube to stabilize the plasma plume. The z-axis was defined along the quartz tube as represented in figure 2. The inter-electrode region of 0 ⩽ z ⩽ 20 mm was the plasma area, whereas the downstream of z ⩾ 30 mm was regarded as the plasma plume.
An Nd:YAG (355 nm)-pumped pulsed dye laser (Scanmate 2C400, Lambda Physik) generated a 451.2 nm beam, which was converted to a 225.6 nm beam by second-harmonic generation. The laser pulse duration τ L was 20 ns. The 225.6 nm laser was introduced to the observed area, and the beam was vertically focused into a sheet shape on the plasma jet by a cylindrical lens with 200 mm focal length. Fluorescence was detected by a photomultiplier tube (R13456, Hamamatsu), and the stray light of the excitation laser was blocked by optical band-pass and high-pass filters (FF01-840/12-25 and FF01-300/LP-25, Semrock). The fluorescence from the excited O was collected by collimating and focusing lenses positioned perpendicularly to both the laser axis and quartz tube axis. The laser beam's horizontal width was 1 mm, and the estimated laser beam waist size in the vertical direction was approximately 5-10 µm. Therefore, the observed volume V O was 1 (laser beam width) × 4 (quartz inner diameter) × 0.005-0.01 (laser beam height) mm 3 . The time-averaged laser energy was measured by a laser power meter (PowerMAX-USB PM3, Coherent). An excitation laser pulse was delayed from a discharge pulse using a delay generator (DG535, Stanford Research Systems), and the time evolution of the TALIF signal after a discharge pulse was measured by changing the time delay τ . The experimental TALIF signals and laser energies were averaged over 512 shots at each measurement condition. The TALIF signal was compensated using the signals at off-resonant of TALIF excitation and without laser irradiation, thereby eliminating the parasitic signals of discharge emission, laser scattering, and fluorescence of the quartz tube (please see equation (1) in [42] for the specific formula).
The spatio-temporal profiles of atomic oxygen were measured by TALIF at different z and τ . Figure 3 illustrates the schematic concept of atomic oxygen production and decay between the repetitive discharge pulses. The production of atomic oxygen by each discharge pulse is defined as ∆[O], and z can be converted as z = vt using the gas flow velocity v. Atomic oxygen is produced by each discharge and consumed during the afterglow phase, and the following discharge pulse superimposes ∆[O] to the residual O density.

Simulation of O behavior
To analyze the spatio-temporal behavior of atomic oxygen, a simple simulation with zero-dimensional time-dependent model was used. The major scheme of the simulation was similar to that used by Ono and Tokuhiro [25]. The simulation was conducted assuming a laminar flow of the working gas in the quartz tube. A radially uniform flow velocity v was assumed in the calculations for simplicity, while the actual flow velocity in the quartz tube has a radially parabolic profile of Hagen-Poiseuille flow.
The simulation in this study differs from that in [25] in terms of the considered reactions and the radical profiles. We , which is an appropriate assumption considering the OH production process by H 2 O dissociation [25]. Axial distributions of ∆[OH] were estimated by interpolating the measured data at similar conditions as in [25].

Linear plots for the ozone interference evaluation
The linearity of the TALIF signal against squared laser energy was checked. In ozone (O 3 ) rich condition, the 226 nm laser induces photo-fragmentation of ozone, which leads to additional production of atomic oxygen as The photo-fragmentation-originated atomic oxygen results in the additional TALIF signal (ozone interference) as [59] where n z is the ozone density and η O , η z are the proportionality factors for atomic oxygen and ozone, respectively. Figure 4 represents TALIF signals against the squared laser energy at [O 2 ] = 1800 ppm, [H 2 O] = 100 ppm, V = 10 kV, z = 10 mm (center of the powered electrodes), τ = ±2 µs (immediately before and after each discharge pulses). The plots in figure 4 demonstrate the linearity of S F against E L 2 , indicating that the ozone interference is negligible in this study when E L ⩽ 0.3 mJ. The following results are obtained with the laser energy below 0.2 mJ.

Atomic oxygen behavior at the plasma region
Atomic oxygen behavior in the plasma region (0 < z < 20 mm) is discussed in this section. Figure 5 represents the time evolution of atomic oxygen density at the dry condition ([H 2 O] ⩽ 100 ppm) at V = 10 kV, z = 10 mm with different O 2 densities. Atomic oxygen is produced within several microseconds after the discharge, and thereafter decreased until the next discharge.

Production phase of atomic oxygen.
The atomic oxygen production process can be evaluated by the analysis of the production phase of 0 < τ < 10 µs. Waskoenig et al estimated the main process of atomic oxygen production in RFplasma jet as electron-impact dissociation and quenching of excited atomic oxygen [26], while the atomic oxygen production process in 1 kHz-plasma jet may be different owing to different electron density and temperature. Figure 6 represents the relative O density in the production phase at various [O 2 ] normalized by maximum and minimum [O] in the range of −5 < τ < 10 µs. The normalized voltage waveform is also plotted in figure 6 to show the timing of τ = 0 s. The O density at [O 2 ] = 275 ppm at close to the discharge (τ < 2 µs) is Table 2. Considered reactions and their rate coefficients at 298 K. The ratios for R4, R6, R10-R14, and R23 are branching ratios. Refer to [43] for calculation of k for three-body reactions.
Reactions k (cm 3 s −1 ) References   not plotted, because the measured TALIF signal was not reliable due to the small signal ratio against the noise of discharge emission. We can discuss the atomic oxygen production processes in terms of the time constant of increase in [O]. Atomic oxygen production in He-O 2 mixture arises from O 2 dissociation by electron-impact or collision with excited helium (He + or He * ). The reactions that can contribute to the atomic oxygen production are listed in table 3. Here, the time constants τ R of primary reactions are calculated as    [60], which can hardly contribute to O production. Therefore we can conclude that the atomic oxygen production does not arise from RP3 to RP4. The Helium metastable He * also has the potential to produce atomic oxygen via RP5 and RP6, whereas we cannot evaluate the O production via RP6 because O − 2 density is not clarified. Naidis simulated the He * density as 1-6 × 10 12 cm −3 in a He-H 2 O plasma jet at 8.4 kHz repetition [61], and Cadot et al measured the He * density as 0.4-3 × 10 13 cm −3 in a pure He 700 ns-pulsed plasma jet at 20 kHz repetition [62]. These orders of density should not contribute to produce atomic oxygen in the order of 10 15 cm −3 , similar to He + . As a consequence, the atomic oxygen in the low-frequency APPJ should be produced by electron-impact dissociation and the following quenching of O( 1 D), similar to the O-production in the RF-plasma jet [26].     Figure 13 in [25]). Although this rough estimation of the time constant may not be accurate as it is on the assumption of the constant OH density, reaction RD5 is qualitatively consistent with the O-decay rate in figure 7, considering that the OH half-life value is in the order of 1 ms [25]. The assumption of a constant OH density is proved to be appropriate in section 3.4.2, considering a cyclic reproduction process. Figure 8 figure 7, indicating the contribution of wateroriginated particles on the decay of atomic oxygen. The aforementioned discussion leads to an important feature that the several ppm of OH predominates the decay of atomic oxygen, rather than several thousand ppm of O 2 . In section 3.4.2, we discuss the reason why OH with 10 13 cm −3 density can affect the behavior of O with 10 15 cm −3 density.

Dependence of O production on the discharge condition
In this section, the dependence of atomic oxygen production in each discharge pulse, ∆[O], in the plasma region (z = 10 mm) on the discharge condition is discussed.

Dependence of ∆[O]
in the discharge area (z = 10 mm) on the applied voltage and discharge energy at dry  condition ([H 2 O] ⩽ 100 ppm) was investigated. Figure 9 represents the current waveforms for different applied voltages V at [O 2 ] = 1000 ppm. The voltage waveform at V = 10 kV is also represented by the dashed line in figure 9. When the applied voltage was increased, the discharge current exhibited faster growth with less duration. This is because the higher instantaneous value of electric field leads to the faster propagation of discharge [65]. The relationship among the applied voltage, discharge energy, and ∆[O] is represented in figure 10. The discharge energy E d was calculated as E d =´V · Idt. According to figure 10, atomic oxygen production ∆[O] is proportional to E d , and its dependence is represented as     of the probable reasons is the balance between the electronimpact dissociation frequency and the number of electrons exceeding the energy threshold of O 2 dissociation. The frequency of electron-impact dissociation increases when [O 2 ] increases, whereas the mean electron energy decreases due to rotational/vibrational O 2 excitation, leading to lower number of electrons contributing to O 2 dissociation. Enhancement of electron attachment by increasing [O 2 ] can cause a decrease in the electron density, which also leads to a similar effect of suppression of O 2 dissociation. As a consequence of these opposing effects, atomic oxygen production can be saturated against From the fitted line in figure 13, The typical O densities in the previous studies of pulsed DBD or RF plasma jets in atmospheric pressure are represented in table 5. Here, SED in table 5 is the specific energy density: SED = (dissipation power)/(flow rate). As represented in table 5, the dissociation rate of O 2 in this study is between those of DBD and RF plasma jets, although a simple comparison is difficult owing to different gas compositions and SED.

Spatial distribution of O density in the downstream of the plasma jet
Considering the application of the plasma jet, the axial distribution of [O] in the downstream of the plasma region (z > 30 mm) is also important. In this section, we analyze the spatial distribution of [O].

Estimation of O density profile by chemical reaction
simulation. Figure 14    cyclically reproduces OH with continuous consumption of atomic oxygen (see figure 16). Therefore, OH reduces atomic oxygen without substantial decrease of itself, like a catalyst. This is why the small amount of OH affects the decay of atomic oxygen with the two-order higher density. This cyclic reproduction of OH is also referred in the literatures considering the OH behavior in the streamer discharge [65,69]. The profiles of [O] and [OH] simulated by full reactions under the same condition as that in figure 15 are represented in figure 17. The density of OH decays immediately after OH production by each discharge (seen as the positive spikes), and thereafter gradually increases or remains constant due to the cyclic reproduction. By contrast, O density has monotonic decreases between discharges. The relatively constant profile of [OH] supports the consideration that OH catalytically consumes atomic oxygen.   is supposed to be significantly high, and the TALIF for [O] b would not be correct owing to large ozone interference (see equation (7)). Therefore, the data at [O 2 ] = 8600 ppm for z > 40 mm are omitted in figure 18. We confirmed the linearity of E L 2 − S F in the plasma region as shown in figure 4, but did not measure in the plume.  figure 19. In contrast to figure 18, the humidity of the working gas strongly  slope is the acceleration of the temporal O decay due to increased OH, H, and HO 2 , therefore this factor is evaluated. According to a previous report using the same experimental equipment, the slope of z- [OH] is not different between [H 2 O] = 100 and 1000 ppm (figure 11 in [25] [25]. The simulated results indicate that the increase in OH leads to only the amplified width of saw-tooth, without changing the slope of z- [O]. Therefore, we can conclude that the change in the z- [

Conclusion
We investigated the behavior of atomic oxygen by TALIF in a 1 kHz-repetition pulsed plasma jet in atmospheric-pressure . Chemical reaction analysis revealed that the cyclic reproduction of OH via chain reaction with O and O 2 is important, therefore a small amount of OH catalytically consumes atomic oxygen with two-order higher density. Humidity in the working gas affected the decay rate of [O] against increasing distance z from the plasma region. This increase in the decay rate of z- [O] slope is attributed to ∆[O] suppression rather than the acceleration of O consumption.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.