Instantaneous imaging of ozone in a gliding arc discharge using photofragmentation laser-induced fluorescence

Ozone vapor, O3, is here visualized in a gliding arc discharge using photofragmentation laser-induced fluorescence. Ozone is imaged by first photodissociating the O3 molecule into an O radical and a vibrationally hot O2 fragment by a pump photon. Thereafter, the vibrationally excited O2 molecule absorbs a second (probe) photon that further transits the O2-molecule to an excited electronic state, and hence, fluorescence from the deexcitation process in the molecule can be detected. Both the photodissociation and excitation processes are achieved within one 248 nm KrF excimer laser pulse that is formed into a laser sheet and the fluorescence is imaged using an intensified CCD camera. The laser-induced signal in the vicinity of the plasma column formed by the gliding arc is confirmed to stem from O3 rather than plasma produced vibrationally hot O2. While both these products can be produced in plasmas a second laser pulse at 266 nm was utilized to separate the pump- from the probe-processes. Such arrangement allowed lifetime studies of vibrationally hot O2, which under these conditions were several orders of magnitude shorter than the lifetime of plasma-produced ozone.


Introduction
Ozone vapor, O 3 , is here visualized in a gliding arc discharge using photofragmentation laser-induced fluorescence. Ozone is imaged by first photodissociating the O 3 molecule into an O radical and a vibrationally hot O 2 fragment by a pump photon. Thereafter, the vibrationally excited O 2 molecule absorbs a second (probe) photon that further transits the O 2 -molecule to an excited electronic state, and hence, fluorescence from the deexcitation process in the molecule can be detected. Both the photodissociation and excitation processes are achieved within one 248 nm KrF excimer laser pulse that is formed into a laser sheet and the fluorescence is imaged using an intensified CCD camera. The laser-induced signal in the vicinity of the plasma column formed by the gliding arc is confirmed to stem from O 3 rather than plasma produced vibrationally hot O 2 . While both these products can be produced in plasmas a second laser pulse at 266 nm was utilized to separate the pump-from the probe-processes. Such arrangement allowed lifetime studies of vibrationally hot O 2 , which under these conditions were several orders of magnitude shorter than the lifetime of plasma-produced ozone.
Keywords: ozone, photofragmentation, laser-induced fluorescence, imaging, plasma, gliding arc (Some figures may appear in colour only in the online journal) Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. efficiency in order to comply with legislation and fulfill the goals defined by environ mental authorities is an important aspect where plasma assisted combustion, PAC, might be an important tool. Plasma has the capability to stabilize flames under leaner conditions as well as extending the explosion limit [8]. Ju et al [8] have listed various areas where PAC has great potential, but also where there are major challenges in the future, for example, the lack of reliable kinetic models in the complex environ ments caused by the plasma.
Numerous investigations have been carried out to increase the knowledge of what conditions can be established with non-thermal plasmas. In a doctoral thesis by Jiajian Zhu several optical diagnostic techniques for various measurements in non-thermal plasmas are discussed [9]. For example, Rayleigh scattering may be used to measure gas temperatures in plasmas. Zhu et al [10] measured the translational temper ature of the hot gas surrounding a plasma column using Rayleigh scattering from a frequency-doubled Nd:YAG laser and found that the translational temperature was roughly 1100 K. Reactive species, such as OH, H, O, and O 3 , are important to measure since they can be utilized in various applications. For example, in plasma-based sterilization the germicidal effect is caused by O 3 and its interaction with cellular respiration [11] and in RF (radio-frequency) plasmas the fuel oxidation process is found to be triggered by O, H, and OH radicals [12]. For further details regarding various optical measurements of reactive species in atmospheric pressure non-thermal plasmas the reader is referred to the review article by Ono [13] and references therein. Also, Aldén et al [14] and Kohse-Höinghoaus [15] have reviewed different laser-based techniques to measure intermediate species relevant for combustion processes.
Imaging of radicals and species distributions using planar laser-induced fluorescence (PLIF) has shown great potential in a variety of research fields including plasma research. Various species can be imaged by forming the laser beam into a sheet and tuning the laser wavelength to match a molecular transition, e.g. OH [16,17] at 283 nm and NO [18] at 226 nm.
Ozone has shown great potential for combustion enhancement since it thermally decomposes into O and O 2 , initiating and accelerating chain-branching reactions, leading to both earlier CH 2 O production and increasing the laminar flame speed [19,20]. As already mentioned, ozone is one reactive species that is created in plasmas and several studies have been carried out to investigate ozone formation in plasmas using UV absorption. Two-dimensional (2D) UV absorption measurements in a streamer discharge have been undertaken by Hegeler et al [21], while Ono et al [22] conducted similar studies in a pulsed corona discharge as well as in a pulsed positive dielectric barrier discharge [23]. These authors have studied both the temporal and spatial distribution of ozone density by either mapping an area through pointwise measurements with a photomultiplier tube, PMT, or by expanding the beam and detecting the absorption with a CCD camera. Although 2D absorption measurements enhance the knowledge of plasma-produced ozone in the spatial domain, UV-absorption suffers from being a line-of-sight technique, for which data interpretation is ambiguous.
In the present work instantaneous 2D distributions of ozone, O 3 , have been imaged in a gliding arc discharge by photofragmentation laser-induced fluorescence (PFLIF). A single laser pulse produced by a KrF excimer laser (248 nm) is used for photodissociation and resonant excitation. Within the pulse a photon first dissociates the O 3 molecule into an O atom and a vibrationally hot O 2 fragment, whereupon a second photon excites the O 2 fragment. The laser excited oxygen molecule then de-excites and emits fluorescence. For further details concerning PFLIF of ozone the reader is referred to [24]. Prior to the ozone measurements the lifetime of vibrationally hot O 2 was measured, both in a laminar ozone flow and in ozone produced by the gliding arc discharge, by letting a 266 nm laser pulse act as photolysis pulse. This arrangement allowed the 248 nm beam to probe the created hot O 2 fragments and the lifetime of hot O 2 could be extracted for both setups.

Experimental setup and measurement concept
The experiments were conducted in three different measurement configurations, case I, case II, and case III, see figure 1. In case I a pump-probe experiment was carried out in a laminar ozone flow. Here, a frequency quadrupled Nd:YAG laser (Quantel, Brilliant b) generated pulses of 266 nm radiation and acted as a photolysis laser. The laser was operating at a pulse repetition rate of 10 Hz with a pulse duration of roughly 5 ns. An injection-locked KrF excimer laser (Lambda Physik EMG 150 MSC), tunable from 247.9-248.9 nm, was used to probe the hot O 2 fragments. The laser was operated at a pulse repetition rate of 10 Hz, producing pulses of roughly 17 ns pulse duration and ~100 mJ pulse energy. The laser linewidth was ~0.2 cm −1 and tuned to an absorption peak, the P(13) line band, of the O 2 fragments. The two laser pulses, i.e. at 266 and 248 nm, were spatially overlapped prior to the measurement volume using a dichroic mirror, highly transmitting for the 266 beam and highly reflective for the 248 nm beam. The laser pulses were then guided through sheet-forming optics before propagating through the measurement volume, containing a well-defined O 3 flow before being terminated in a beam dump. The O 3 flow was generated by supplying oxygen through an ozone generator (O 3 -Technology AC-20), and then issued into the probe volume through a Teflon pipe. The oxygen gas flow was controlled using mass flow controllers (Bronkhorst). The massflow rate of ozone could be varied either by varying the flow of oxygen or the efficiency of the ozone generator.
An intensified CCD camera, ICCD (Princeton Instruments, PIMAX III), was used to capture the O 2 emission. The ICCD camera was equipped with an 85 mm focal length UV lens and two long-pass filters, WG335 and a liquid n-butyl acetate filter, in front of the camera and thereby wavelengths longer than 320 nm were transmitted to the ICCD camera. A spectrograph (Shamrock 750i, Andor, 1200 grooves mm −1 grating) mounted with an ICCD camera (iStar, Andor) was employed in order to capture the emission spectra from excited O 2 fragments between 320 and 450 nm. The collection optics in front of the spectrograph consisted of two spherical lenses with focal lengths, L2 = 300 mm and L1 = 200 mm, respectively. The fluence of the two laser beams in the probe volume were 0.03 J cm −2 for the 248 nm pulse and 0.1 J cm −2 for the 266 nm pulse. A pulse generator (BNC 575) provided pulses to synchronize the two lasers and also enabled variation of the time delay between the two laser beams. The measurement was carried out 3 cm above the output of the Teflon pipe providing the generated ozone flow.
In case II, the measurements were conducted in a gliding arc setup, see figure 1. Here, a plasma was created by the gliding arc discharge system, which was set in open air. It primarily consists of three components: electrodes, a power supply, and an air flow. The hollow tube-shaped electrodes are made of stainless steel, having an outer diameter of 3 mm and are internally water-cooled. The gliding arc discharge was driven by an AC power supply (Generator 6030, SOFTAL Electronic GmbH, Germany). For further details concerning the gliding arc discharge system the reader is referred to [17]. The pulse generator provided pulses to synchronize the two lasers with the plasma, and also enabled variation of the time delay between the two laser beams. The input power to the gliding arc was 1000 W and the air flow was kept at 15 l min −1 . Except for the ozone flow and the spectrograph in case I, the same equipment and laser fluences were used in case II. Finally, a third set of measurements were conducted, case III, where only the excimer laser was operating. This single-pulse configuration allowed the 248 nm laser pulse to act as both pump and probe pulse. A spectrograph was employed in order to capture the emission spectrum from excited O 2 fragments with the same configuration as in case I. Except for the lack of the 266 nm pulse, case III was operated with the same equipment and fluences as case II. For cases II and III the measurements were conducted 9 cm above the Teflon based plate which fixated the electrodes, see figure 2. In figure 2, the size and position of the 248 nm laser beam is shown as the horizontal blue line.

Results and discussion
Ozone was generated into the probe volume either by a laminar ozone flow or produced by the gliding arc discharge. For the lifetime measurements in cases II and III the gliding arc discharge was turned off prior to the arrival of the pump pulse to ensure that plasma-produced hot O 2 was negligible in the probe volume.

Pump-probe delay study in a laminar ozone flow
In plasmas both O 3 and hot O 2 are produced, wherefore an initial study was performed in order to measure the lifetime of hot O 2 , see case I in figure 1. At ambient conditions O 3 has a lifetime of 30-40 min [19], whereas hot O 2 is expected to be short-lived before relaxing to a thermal equilibrium through  fragments were therefore separated by two different laser pulses, 266 and 248 nm, respectively. Here, the 266 nm laser pulse acts as a pump-pulse and dissociates the ozone molecule into an oxygen atom and a vibrationally excited O 2 fragment, while the 248 nm laser pulse acts as a probe-pulse and excites the hot O 2 distribution to a higher electronic state. By varying the time delay between the pump and probe pulses, the lifetime of hot O 2 was determined. Figure 3 shows the detected O 2 fluorescence signal at different time delays of the probe pulse after dissociation of ozone. At t = 0 the pump pulse was fired and the ozone dissociated. Each measurement point is an average of 10 software (SW) accumulations each covering 50 hardware (HW) accumulations, which were performed in a gas flow containing roughly 4% ozone. A background image, where no lasers were fired, has been subtracted for all data points. The increase in signal during the first 100 ns is most probably a result of population of higher vibrational states than the O 2 (X 3 Σ − g , ν = 6, 7) after photolysis of ozone. The de-excitation of these states may lead to an initial increased population of the ν ″ = 6,7 states which results in an increase of the fluorescence signal. Similar results have been reported for ν ″ = 8,9 [25] upon 266 nm photolysis, in which vibrational population up to ν ″ = 22 is energetically possible. From this study a lifetime of 550 ns is obtained for vibrationally hot O 2 in the ozone flow. A spectroscopic investigation was conducted in order to verify that the detected signal was stemming from hot oxygen. It was found that the dispersed emission spectrum coincides with emission originating from vibrationally hot O 2 [24].

Pump-probe delay study in gliding arc
In a similar manner, as for the results presented in figure 3, the lifetime of hot O 2 was also investigated for ozone produced by the gliding arc discharge. The 266 nm laser pulse is used to dissociate ozone and the 248 nm laser pulse to probe the hot O 2 fragments at different time delays after dissociation, see case II in figure 1. The measurements were conducted 0.5 ms after the gliding arc discharge was turned off to ensure that the electron density was too low to produce hot O 2 [26]. The result is depicted in figure 4, where each measurement point is an average of 10 SW accumulations, each covering 50 HW accumulations. Each data point has been compensated for background recorded with no lasers fired. The input power to the gliding arc was 1000 W and the air flow was kept at 15 l min −1 . Due to the random fluctuations of the gliding arc discharge, the absolute value of the lifetime for the hot O 2 fluorescence signal is difficult to determine, but has a similar magnitude as the measurements performed in the ozone gas flow, 90 ns (±40 ns). From 20 to roughly 1000 ns there is a distinct signal drop, whereas for delay times longer than 1000 ns the signal is rather constant. This signal background is a result of self-photolysis from the 248 nm pulse, i.e. this signal is independent of the 266 nm pulse and the 248 nm pulse acts both as pump and probe pulse. The signal fluctuations are a result of the interaction between the turbulent gas flow and the transient gliding arc plasma. The location of the signal in the images recorded for each data point is essentially unchanged regardless of the time delay, indicating that the influence of the gas flow is minor. Potential signal interference from chemically and electronically produced hot O 2 is discussed below.

Lifetime of ozone produced by the gliding arc discharge
The lifetime of ozone generated by the gliding arc discharge was investigated for delay times between 0.5 and 5 ms after the discharge was turned off, i.e. case III in figure 1, and the result is shown in figure 5. For this study, only the 248 nm laser was operating, thus acting as both pump and probe laser. The discharge was turned off at t = 0 and the signal decay reveals the consumption of the plasma-produced ozone over  time. A signal lifetime of 1.9 ms is retrieved from the graph, where each measurement point is an average of 20 SW accumulations, each containing 50 HW accumulations. Each data point has been compensated for background recorded with the same settings but without firing any laser pulse. The influence of the gas flow on the data points is minor since the signal distribution in the images recorded for each data point is essentially unaffected by the time delay. A spectrograph was used to ensure that the detected emission was stemming from hot O 2 . The ozone has a lifetime on the order of milliseconds, which compared to the lifetime of hot O 2 (~90 ns) is roughly four orders of magnitude longer. Thus, imaging of ozone using PFLIF with a 248 nm laser pulse is possible if the laser pulse is fired when the electron density is low enough to ensure that electron collisions will not put significant population in ν″ = 6,7 of O 2 . The electron density is, typically, low enough within tens of nanoseconds after turning off the high voltage [26]. Although, chemical reactions that produce hot O 2 will occur after the gliding arc has been shut off, the probability that these reactions will put significant populations in O 2 (ν″ = 6,7) decreases rapidly after the discharge is shut off and after 0.5 ms the population is assumed to be negligibly small.

Electronically versus photochemically produced hot O 2
The one-pulse photofragmentation approach has a potential challenge in discriminating signal that stems from ozone and plasma-produced vibrationally excited oxygen (ν″ = 6,7) in a plasma. Such vibrationally excited oxygen could potentially be a result of electronic collision excitation. Previously presented data indicate that the kinetic energy of the electrons and the ambient gas temperature in the gliding arc in the current study is around 0.8 eV and in the range of 1100-1600 K, respectively [10]. An oxygen molecule has to be in at least ν″ = 2 to gain enough energy through an electronic collision to reach ν″ = 6 [27]. At gas temperatures around 1100-1600 K, the fraction of O 2 that are in ν″ ⩾ 2 is typically 1.5%-6% assuming a Boltzmann distribution. Under these conditions, typical cross section values for such electronic-collision excitations are on the order of 10 −20 cm 2 (retrieved from extrapolation of table 5 in [28]). Given that the laser pulse duration was 17 ns, the pulse energy 40 mJ, N O3 = 1000 ppm, N e = 7.3 · 10 18 m −3 [29], the number of hot O 2 (ν″ = 6,7) that are created from electronic collisions, in this short time interval, was calculated to be roughly five orders of magnitude less than the hot O 2 (ν″ = 6,7) fragments created by laser fragmentation of O 3 . This large difference is a result of the hot O 2 being created from electronic collisions which is continuously produced over time by the gliding arc discharge, while all hot O 2 fragments produced by PFLIF of O 3 is created only during the 17 ns laser pulse duration. We therefore conclude that ozone imaging is possible when the gliding arc discharge is operating since any signal interference from hot O 2 produced by electronic collisions is negligible for this setup. It should, however, be emphasized that the signal interference from plasma-produced hot O 2 could vary for different plasma sources and must therefore be considered for the investigated plasma source to ensure accurate ozone visualization.

Imaging
The single-shot images shown in figure 6 display the 2D distribution of the O 2 fluorescence signal based on PFLIF of ozone, using 248 nm laser radiation, in the plasma produced by the gliding arc discharge, see case III. Here, figures 6(a) and (b) are images recorded when the discharge was on, while figure 6(c) is acquired 250 µs after the discharge was turned off. The air flow was kept at 15 l min −1 for all images, whereas the input power to the gliding arc was 600, 400, and 800 W for 6(a)-(c), respectively. A short camera gate, 50 ns, was kept for the result shown in figures 6(a) and (c), while the camera exposure time was long enough, 500 ns, for the data shown in figure 6(b) so that the plasma column could be clearly seen as the small string in the bottom center. The spatial resolution for all images in figure 6 is 0.043 mm/pixel. Furthermore, based on the aforementioned calculations it was concluded for this setup that the number density of naturally present vibrationally hot O 2 in the plasma was several orders of magnitude lower than the number density of vibrationally hot O 2 stemming from photolysis of O 3 and hence the plasma did not have to be turned off in order to measure O 3 . The fluorescence from hot O 2 thus stems from O 3 whether the plasma is turned off or continuously running. Figure 6(b) shows the spatial distribution of ozone in the vicinity of the discharge channel. Ozone has a distinct distribution characteristic compared with the distribution of OH, which has been reported in a previous work by Zhu et al [30]. They found a hollow structure for OH, which is not observed for O 3 in this work. Since ozone has a lifetime on the order of milliseconds, in the gliding arc setup, it is not surprising that O 3 is found in a region outside the plasma column, as can be seen in figure 6(b). The signal is wide compared to the plasma column, which also is a result of the long lifetime of ozone. In addition, the vibrationally excited O 2 molecules in ν″ = 6,7 and electronically excited O 2 (in the B 3 Σ − u state) is unaffected by collisional The gliding arc discharge has been turned off at t = 0 and the red line is a single exponential fit to the data points.
quenching, since the vibrational energy transfer process is much slower than the laser pulse duration and predissociation is the dominating loss factor of the electronically excited state [24].

Summary and conclusion
In the presented study, a single laser beam from a KrF excimer laser was fired to visualize ozone in a gliding arc discharge. The concept is based on a pump-probe technique called PFLIF, where the 248 nm laser pulse acts both as pump and probe pulse.
In this one-pulse configuration the pump pulse dissociates O 3 into O and vibrationally hot O 2 fragments, while the probe pulse excites the created hot O 2 fragment. Imaging of O 3 in the plasma must be carried out with caution since vibrationally hot O 2 can be produced by the plasma itself. The validity of applying this method to monitor ozone in a gliding arc was investigated by adding an additional 266 nm preceding laser pulse that acts as pump pulse, whereas the 248 nm pulse now mainly acts as a probe pulse. With this two-pulse configuration the lifetime of hot O 2 could be estimated both in a laminar O 3 flow and in the gliding arc discharge. It was found that the lifetime of hot O 2 is several orders of magnitude shorter than the lifetime of O 3 produced by the plasma, which ensures that O 3 can be monitored when the plasma is turned off. In addition, the amount of plasma-produced vibrationally hot O 2 was estimated to be several orders of magnitude lower than the amount of hot O 2 stemming from O 3 , which enable ozone imaging with this method when the gliding arc plasma is present.