Single-electrode He microplasma jets driven by nanosecond voltage pulses

The recent development of non-equilibrium plasma jets for biomedical applications brings refreshed interest both in the physics of guided streamers^1–3 and in the biochemistry associated with the plasma-induced oxidation-reduction.^4,5 These biomedical applications ranged from wound healing,^6–10 infectious disease control,^8,11–16 cancer therapy^17–19 to dental applications.^20,21 More recently, the non-thermal plasma jets were also considered as novel approaches for surface and material processing.^22,23 The unique feature of plasma jets–the plasma plume escaping from both electrodes and reaching area that are not constrained by physical location or geometry of the electrodes–makes them attractive for a variety of biomedical and environmental applications in addition to their non-equilibrium property similar as many other non-equilibrium plasmas, i.e., the electrons of the plasma are energetic and produce reactive plasma species with enhanced ionization, dissociation, and excitation, while the temperature of the heavy particles remains near room temperature. The length of the plasma jets or the maximum distance between the launched ionization wavefront and the nearest dielectric or electrode surface ranged from micrometers to 11 cm.^12,24–29 However, the reported effective sterilization or penetration depth is no more than 1 cm,³⁰ which may be due to the significantly reduced production of reactive plasma species such as O at the distances away (e.g., >5 mm) from the powered electrode.^31,32 This reduced sterilization effect along with limited accessibility of millimeter-diameter plasma jets into complex structures have become serious hurdles for biomedical applications of non-equilibrium plasmas, particularly in areas where complete sterilization of channels or volumes with millimeter or less openings such as endoscopes or root canals is crucial to the treatment success.³⁰ Lu et al. proposed to use a hollow needle as the single electrode, powered by microsecond high voltage pulses or RC-filtered microsecond high voltage pulses, for root canal disinfection.³³ As this simple configuration-based plasma generation is very interesting to biomedical applications,³⁴ studies on the plasma jets, however, are only a few^35,36 and were limited to relatively longer high voltage pulses with rise times longer than 100 ns. It was reported that both the length of plasma plumes and the maximum plume propagation velocities increased when the rise time of a 5-μs-duration pulsed voltage decreased from 4 μs to 100 ns for a dielectric tube-contained Ar plasma jet.³⁷ The effects of rise time of the voltage pulses on the dynamics of a guided streamer were studied with simulations using a fluid model for a He plasma jet in air excited by pulsed voltages with rise times ranging from 500 ns to 50 ns.¹ However, so far there is no study of plasma dynamics for atmospheric pressure plasma jets excited with shorter rise times, e.g., less than 40 ns, possibly due to the limit supply of nanosecond high voltage pulsed power supplies as well as the technical challenge in efficient coupling of ultrashort pulses to the plasma source.

We present here a single-electrode microplasma jet that is excited by 5 ns, 8 kV pulses at a single-shot or 500 Hz. Dynamics of the 5 ns pulsed plasma jet was compared with the ones excited with a longer voltage pulse, i.e., 140 ns, at the same maximum voltage. High speed imaging was conducted to assess the plume formation and the propagation velocity of the ionization wavefronts for the two plasmas excited with two different nanosecond high voltage pulses. Generation of electronically excited species including O(⁵P), N₂(C³Π_u), and $N2+(B2Σu+)$ were compared using optical emission spectroscopy. In addition, the rotational and vibrational temperatures were measured for both microplasma jets.

The single electrode was made of a 25-gauge stainless steel syringe needle (0.51 mm outer diameter and 0.26 mm inner diameter) with a shovel tip (Monoject^® Hypodemic needles). Helium was used as the working gas and flew through the hollow needle at a flow rate of 500 sccm (standard cubic cm per min), which was controlled by a mass flow controller (MKS Instruments 2179). The length of the plasma jet depends both on the discharge voltage and on the gas flow rate. At a given voltage, e.g., 8 kV, a range of flow rates between 200 sccm and 1000 sccm were tested. A helium flow at 500 sccm was able to achieve the maximum plasma jet length with stable laminar flow at 8 kV and was thereby used for the comparison study.

Nanosecond high voltage pulsed power sources were provided by the Pulsed Power Group at the University of Southern California. An ultrashort pulse generator produces 5 ns-FWHM (full width at half maximum), 4–10 kV pulses at a single shot or up to 500 Hz (for the pulse repetition rate). Based on magnetic pulse compression and diode opening switching, this compact high voltage pulse generator is capable of providing rise times less than 5 ns into a 50 Ω load.³⁸ For a high impedance load such as the weakly ionized atmospheric pressure plasma, the maximum voltage output tested was 10 kV with a rise time of 5 ns when the power supply was externally triggered at a repetition rate of 500 Hz. The external triggering jitter of the pulsed power supply is relatively high, in the range of 5–10 ns. For better synchronization, an attenuated voltage directly from the pulsed output was used to determine the actual timing between the discharge, the camera, and the intensifier gating. For comparison, a high voltage pulsed power supply providing longer duration pulses was also used to drive the single-electrode plasma source. This inductive adder-based pulse generator outputs 200 ns, ≤10 kV pulses at a single shot or up to 5 kHz with the triggering jitter less than 1 ns. The rise time of 8 kV voltage pulses on the plasma load was measured to be 60 ns.

Diagram of the experimental setup for fast imaging of the plasma jet is shown in Figure 1. The setup consists of the single-electrode plasma source, the pulsed power supply, an intensified CCD (ICCD), a function generator (Stanford Research Systems DG535) for synchronization, and a customized V-I monitor coupled with a fast digital oscilloscope for voltage and current measurements. The electrode is connected to the inner conductor of a 3-ft-long 50-Ω coaxial cable (Pasternack, RG213), which functions as a transmission line and delivers the voltage pulses from the pulsed power supply to the plasma load, as shown in Figure 1. The ground of the power supply is connected with one end of the outer conductor of the coaxial cable. The other end of the outer conductor of the coaxial cable is left open. The V-I monitor, inserted between the coaxial line and the load, consists of a current probe (Pearson 6585) and a resistive divider and allows broadband measurements, e.g., current rise time of 1.5 ns and bandwidth up to 1.5 GHz.³⁹ The resistive divider consists of two low-inductive resistors in series and serves as a voltage probe with the attenuation of 1:1000. Both the voltage and current probes were connected with the oscilloscope using the same length coaxial cables for power measurements. The ICCD camera system (Princeton Instruments PI-MAX3, Gen II) is equipped with a fast-gate intensifier allowing subnanosecond synchronization and coupled with a lens system (Nikon Micro-Nikkor 105 mm f/2.8) to collect emission from the nanosecond microplasma jet.

For optical emission spectroscopy, a spectrograph (Acton SP-2758) with visible-optimized grating (1800 G/mm) was coupled with the ICCD camera to collect the emission spectra. Two UV enhanced metallic aluminum mirrors, a plano mirror and a focusing mirror (focal length = 75 mm), were used to collect the optical emission from the plasma jets. The magnification of the optical system was set at one, which produced a 1:1 inverted image of the jet. Optical emission was collected for the wavelength range of 200 nm to 850 nm and was integrated for 4000 voltage pulses. Additionally, a longpass filter (Thorlabs FEL0450) was inserted at the spectrometer slit when measuring wavelengths above 500 nm to block any interference due to second order dispersions from the radiating species below 450 nm.

The discharge voltage and current waveforms are shown in Figures 2(a) and 2(b) for excitations with shorter and longer nanosecond voltage pulses, respectively. The FWHM of the shorter voltage pulse at 8 kV was measured to be 4.6 ns and the rise time 5.1 ns. Relatively small negative peaks, about 30% of the positive maximum, for the voltage pulses were observed due to reflection between the load and the pulsed power source. A pulsed current containing both displacement and conduction components were measured and the maximum positive peak reached 28 A for the application of 5 ns, 8 kV pulses. In comparison, the longer voltage pulses induced a primary positive peak with a pulse width of 140 ns and a rise time of 60 ns at the electrode. The corresponding current pulse has its positive maximum of only 700 mA, smaller than the peak current excited with shorter pulses by a factor of 40. Although both the current pulses were dominated by their displacement components, the current and voltage pulses have more in-phase portion for the shorter pulses compared with the longer ones. The phase difference between the shorter and longer pulses implies impedance difference in the plasma, which served as a load for the transmission line of the voltage and current pulses. Both the higher peak current (due to higher dV/dt) and the increased in-phase portion resulted in slightly higher energy delivery to the plasma load for the shorter excitation pulse: about 81 μJ per pulse for the 5 ns and 70 μJ per pulse for the 140 ns pulses at the same amplitude of 8 kV. The energy per pulse was calculated by applying integration of the product of voltage and current pulses over a sufficient time period, e.g., 50 ns for 5 ns pulses and 1 μs for 140 ns pulses.

To understand the dynamics of the plasma jets and the impact of applied voltage pulses, temporal development of single-electrode microplasma jets excited with 5 ns and 140 ns voltage pulses was studied. For the 5 ns pulsed plasma jet, time-sequenced images were obtained starting from the onset of the plasma plume till the afterglow emission, as shown in Figures 3(b)–3(m). The delay time, as indicated at the bottom of each image, was defined as the time after the onset of the plume from the electrode tip. The zero delay time, i.e., t = 0, corresponds to the time when the discharge voltage reaches about 3 kV and the discharge current starts, as shown in Figure 2(a). To maximize the signal-to-noise ratio, each image in Figure 3, except Figures 3(a) and 3(n), was obtained by integrating 1000 images at the given delay time with a constant ICCD gate width (exposure time) of 3 ns. For the first 30 ns, the images were taken at every 5 ns. After 40 ns, the time interval between images was increased to 20 ns. In Figure 3, selected images after 40 ns were shown to illustrate the changes in the plume development. The plasma plume reached its full length of 8 mm within the first 30 ns, after which the plasma plume started decay and the visible emission was detectable till 240 ns. Note that the spatial distance z = 0 was set at the tip of the needle, as shown in Figure 3(a) for the needle setup. Figures 3(c)–3(l) showed that the plasma plume extended to the top edge of the shovel needle, where z < 0, after the discharge initiation. In addition, the plasma emission expanded radially (<1 mm radius) at the needle tip (Figure 3(c)) within the first 5 ns. This radial emission decayed quickly within 20 ns, while the plasma plume was still growing axially up to 30 ns. The observed radial expansion of the plasma followed with a rapid decay implied that the applied ultra-short pulsed electric field reached its maximum at the electrode tip and its adjacent areas within the first 5 ns and were responsible for the initial breakdown and the ionization process of the jet. As the helium gas flowed axially, the “guided” ionization wavefront formed rapidly along the axis and away from the tip of the electrode. At the radial direction, the plasma plume grew outward from the surface of the electrode tip due to the radial component of the pulsed electric field and was not able to propagate further away radially because of decrease of the amount of helium as a guided buffer. Figure 3(n) shows a single-shot image with the exposure time of 100 ns, where the radial emission was too weak to detect comparing to the axial emission, especially when the orifice of the shovel-tipped needle was turned to face backwards, and the total axial length of the plasma jet appeared shorter due to the limited sensitivity of the detector.

For the 140 ns pulsed plasma jet, the ionization channel formed between the streamer head and the needle electrode appeared continuous within the first 40 ns (Figures 4(b)–4(g)). A dark space started forming behind the streamer head after 50 ns while the streamer head grew both axially and radially with time, which resulted in a mushroom-like shape, as shown in the integrated images based on 1000 exposures (Figures 4(h)–4(l)). Note that the dark space or the less luminous region formed behind the streamer head was similar to the dark channels observed in the longer pulsed plasma jets,^40,41 and its conductivity may be sufficient to impact the propagation of the streamer head.⁴¹ The ionization wavefront reached z = 18 mm from the nozzle tip and its radius grew to be 3.5 mm at 100 ns (Figure 4(l)). However, the single-shot image (Figure 4(m)) revealed that the individual discharge event started as a guided-streamer and transitioned to a stochastic branching. The appearing radial expansion of the streamer head observed in the integrated images was in fact occurrence of branching, in which the streamer grew stochastically in both radial and axial directions, similar to a positive air streamer.^42–44 This transition from a guided streamer to a stochastic branched streamer may be due to the loss of a guided helium channel with the increasing distance between the streamer head and the electrode tip. In the mean time, the induced electric field at the streamer head was still sufficient to sustain photoionization but in a helium and air mixture. Interestingly, although the individual branching appeared random, the position of the multiple streamer heads at a given delay time was repeatable, exhibiting a reproducible group behavior shown by the integrated images. The emission of the mushroom streamer head decayed rapidly and was barely observable by 110 ns.

Velocity of the ionization wavefront of the plasma jet at the axial direction can be calculated based on the positions of the forefront of the streamer head at given delay times. Figure 5 compares that axial velocity of the ionization wavefront of the plasmas excited, respectively, with 5 ns and 140 ns pulses at 8 kV. A velocity of 8.0 × 10⁵m/s was measured within the first 5 ns for the 5 ns pulsed plasma jet, and it is 2.7 times higher than the highest value, measured at 25 ns, for the 164 ns pulsed plasma jet, as shown in Figure 5. This indicates that significantly higher plume acceleration, about 13 times higher, was obtained for the 5 ns pulsed plasma jet compared with the longer pulsed one. Also, notice that the highest velocities for both cases happened during the voltage pulses or the pulsed electric field in present at the electrode. This implies that the acceleration of the ionization wavefronts is electric field-dependent. Shorter pulses, usually associated with higher voltage increasing rates (i.e., higher dV/dt), favor efficient photon ionization processes in the streamer. In addition, there was a second peak initiated after 70 ns in the axial velocity temporal development for the 140 ns pulsed plasma, which indicated a smaller second acceleration occurred away from the electrode. This may be due to the electric field enhanced by the accumulated space charges at the guided streamer head before the stochastic branching, as in a typical positive streamer in air, dominates.

UV-visible emission spectroscopy was conducted to gain more information about the electronically excited plasma species in the plasma plumes. Figure 6 compares the optical emission spectra ranging from 200 nm to 850 nm for the plasmas excited by 5 ns, 8 kV pulses (Figure 6(a)) and by 140 ns, 8 kV pulses (Figure 6(b)). Transitions including N₂(C-B), N₂⁺(B-X), He(D-P), O(P-S), and OH(A-X) were observed in both plasma emission spectra. Comparing the emission lines of the 5 ns pulsed plasma with the 140 ns pulsed plasma, the intensities of N₂ emission lines were comparable or slightly lower, in the range of 0.8–1, whereas both the N₂⁺ emissions (at 391.1 nm, 427.8 nm, and 470.9 nm) and O emissions (at 777 nm and 844.6 nm) increased by the factor of 1.3 and 1.4, respectively. OH emission lines at 307.8 nm, 308.2 nm, and 309 nm were weak and comparable for both plasmas.

The rotational and vibrational temperatures along the axial center of the plasma jets were determined by fitting the measured emission spectra of N₂ second positive system with the simulated ones using Specair. The experimentally measured emission spectra, as shown in Figure 7, were the line-integration of the emission spectra in the axial direction over the entire length of the plasma jet and thereby reflected the average temperature of the plasma at the axial center. The rotational temperature of the plasma remained at 300 K ± 50 K for both the 5 ns and 140 ns pulsed plasmas. Figure 7(a) shows an example of the rotational temperature measurement for the 5 ns pulsed plasma. Fitting the measured rovibronic transitions of N₂ (C³Π_u − B³Π_g) indicated that the vibrational temperature was 3200 K ± 300 K for the 5 ns pulsed plasma (Figure 7(b)) and 3100 K ± 300 K for the 140 ns pulsed plasma. In addition, spatially resolved vibrational temperature at the axial center, at a spatial resolution of 250 μm, was evaluated for both plasma jets for a distance of 2 mm from the tip of the needle electrode. Within the short distance, the variation of the vibrational temperature was small, about 300 K.

The formation criterion of a streamer is considered as a condition when an avalanche is forming such that the external electric field $E0$ approximates to the field $E′$ induced by the space charge of the streamer head, i.e., $E0≈E′=eR−2 exp (αx)$⁠,⁴⁵ where e is the electron charge, R is the streamer head radius, α is the ionization coefficient and is function of the external electric field, and x is typically considered as the gap distance between the electrodes, but can also be regarded as the shortest distance at which the streamer forms. This was simplified by the Meek breakdown criterion, where the product of the ionization coefficient α and the gap distance d demands $α(E0)d≈18−20$⁠.⁴⁵ Application of voltage with sufficient high rising rate (high dV/dt) allows higher electric field in present during breakdown and causes higher ionization coefficient as well as higher drift velocity of the electrons. The former would result in higher number of electrons generated at the initiation of the discharge for the same characteristic distance d, and the latter would result in accelerated charge separation at the streamer head. For example, an increase of the reduced electric field E₀/p, where p is the gas pressure (760 Torr in our case), by a factor of 2 would result in a factor of 1.4 increase in α/p if we assume the breakdown E₀/p ≈ 100 V/(cm Torr).^45,46 As E₀/p increases, the dependence of α/p on E₀/p decreases; the same factor of increase in the reduced electric field would result in less increase in α/p at higher value of E₀/p. Yet, any slight increase in the ionization coefficient can significantly impact the electron number density during the formation process, as $Ne∼ exp(αd)$⁴⁵ and a factor of 1.04 increase in α would result in one order of magnitude increase in the electron number density. In addition, as the drift velocity is dependent on the external electric field, i.e., $υd=μeE0$⁠, where $μe$ is the electron mobility, the increased drift velocity due to the increase external electric field would expedite the charge separation and leave the streamer head positively charged at a shorter time before the charge losses due to various mechanisms such as recombination and radiation take place. Both the increased electron density and the accelerated charge separation would increase the induced electric field $E′$ at the streamer head and thereby accelerate the initiation process of the streamer. This may have resulted in the significantly increased acceleration, by a factor of 12, of the streamer head in the 5 ns pulsed plasma jets compared with the 140 ns pulsed plasmas. Both the increased current peak and the energy deposition in the plasma pulsed by shorter pulses also support the hypothesis that higher dV/dt increases the ionization coefficient and the electron drift velocity, and also agrees with the simulation findings reported by Boeuf et al. for a He plasma jet in air excited by 4 kV, 1 μs–duration pulsed voltages with varying rise times ranging from 500 ns to 50 ns.¹ 

Nevertheless, higher energy for the discharge initiation or triggering, due to the application of higher dV/dt or shorter voltage pulse rise time, does not necessary mean higher energy for the streamer propagation.¹ Energy is needed for continuous streamer formation and propagation and can only be obtained from the external electric field.⁴⁵ The streamer head in the 140 ns pulsed plasmas reached further distance, by a factor of 2.25, compared to that in the 5 ns pulsed plasma jets. In addition, the different streamer propagation behaviors, exhibited by the plasma jets excited with voltage pulses of different pulse durations, implied that the streamer propagation was strongly dependent on the duration of the electric field at the electrode. After the onset of the plasma plumes, the electric field needed to sustain the discharge or to accelerate ionization wavefronts disappeared after 5–8 ns for the shorter pulses, whereas it lasted for almost 150 ns for the longer ones. The electric field near the electrode with the short duration was able to accelerate the plasma plume to a higher velocity but decayed rapidly before it was able to assist the propagation of the space-charge dominated streamer head to either longer distance or transition into a branching mode, as demonstrated by the plasma jets excited with longer duration pulses. The finding that the length of the plasma jets depends on the duration of the electric field agrees with those found in the helium plasma jets generated by two-electrode systems excited with voltage pulses with longer durations of 200 ns–900 ns.⁴⁷ However, the plume propagation behaviors observed in this study are unique due to both the short duration of electric fields at the electrode and the single electrode configuration without any dielectrics or ground electrode applied nearby.

In addition, the difference in the optical emission spectral distribution between the two pulsed plasma jets indicated different production rates of the electronically excited plasma species for the two types of nanosecond plasma jets, although the rotational and vibration temperatures of the plasmas were comparable. 5 ns pulsed plasma jets favored the production of $N2+(B2Σu+)$ and O(⁵P) but produced less N₂(C³Π_u) compared with the 140 ns pulsed plasmas. Considering the ionization energy required to produce N₂⁺ via direct electron impact (e + N₂ → N₂⁺+ e + e) is 15.6 eV,⁴⁸ whereas the highest electron impact cross sections for rovibrational excitation of N₂ are between 2.1 and 3.0 eV,^49,50 the enhanced production of energetic particles such as $N2+(B2Σu+)$ accompanied with the reduced yield of N₂(C³Π_u) resulted in an increased emission intensity ratio of $N2+(B2Σu+)$ to N₂(C³Π_u) in the 5 ns pulsed plasma. It suggested that a shorter duration voltage pulse excitation of the plasma may have caused a shift of the non-equilibrium electron energy distribution function (EEDF) of the plasma to higher energies, which may have favored the production of the reaction oxygen species, i.e., O(⁵P), compared with that of the plasma excited by longer voltage pulses, as indicated in the results.

Single-electrode helium microplasma jets were generated in ambient air when the electrode was excited with nanosecond 8 kV pulses at 500 Hz. Excitation of the single electrode with 5 ns, 8 kV pulses with a voltage increase rate of 1.28 × 10¹²V/s resulted in fast acceleration of the streamer head that reached a maximum velocity of 8 × 10⁵m/s within 5 ns. This velocity of the streamer head is almost 3 times that of the streamers excited with 140 ns voltage pulses at the same amplitude or of other reported helium jets driven by high voltage pulses with longer pulse durations. It was considered that the fast voltage rising rate enabled higher electric field during the streamer initiation with increased ionization coefficient and electron drift velocity, which increased the energy deposition in the plasma and resulted in the acceleration of the streamer head. In addition, application of high voltage pulses with the short voltage rise time and pulse duration favored the production of reactive plasma species and resulted in enhanced $N2+(B2Σu+)$ and O(⁵P) production. Nevertheless, the role of the voltage rise time needs to be more systematically studied to separate the effects due to pulse durations. Numerical modeling that includes the streamer formation for the single electrode configuration driven by pulsed voltages with ultra-short rise times and durations are also needed to quantify and better understand the observed phenomena and effects.

This material is based upon work supported by the Air Force Office of Scientific Research under AFOSR Award No. FA9550-11-1-0190.