Introduction

Atmospheric pressure nonequilibrium plasmas recently have attracted a great interest largely because it can be used for a wide range of applications, such as surface modification, thin film deposition and biological decontamination, without the need for an expensive vacuum system1,2,3,4. Such plasmas are usually produced by different kinds of discharges like corona, capillary, direct-current (DC), radio frequency, microwave and dielectric-barrier discharges5. DC glow discharge is a popular one among them, with the voltage applied between two naked electrodes. The DC glow discharge was first observed with helium, neon and nitrogen used as working gas in a tube at the low pressure and characterized by different regions depending on the discharge parameters (e.g., the geometry of the electrodes, the gas type and the pressure)6. This kind of discharge was also successfully generated at high pressure, even at atmospheric pressure in the ambient air, though there exist undesirable instabilities which lead a glow to arc transition. Different geometries of electrode and different types of gas were employed to characterize the DC glow discharge in laboratory. A DC microhollow cathode discharge (MHCD) in atmospheric pressure air has been reported by Stark and Schoenbach7,8. The stability of this plasma is attributed to the fact that the microhollow cathode supplies electrons to the positive column of the discharge rather than the instability prone cathode region as in a traditional glow discharge. Another DC glow discharge was generated between a thin cylindrical anode and a flat cathode with helium, argon, hydrogen, nitrogen and air used as working gas respectively9. The discharges appeared to operate as temperature and pressure scaled versions of low pressure discharges. It is possible to distinguish negative glow, Faraday dark space and positive column regions of the discharge in the normal glow discharges. Investigations of atmospheric pressure glow discharges in helium, argon, nitrogen and air in the similar electrode configuration were presented10. Experiment results showed that for all the gases, plasma non-equilibrium degree in a positive column decreases with discharge current increase.

It is generally accepted that a high speed gas flow is beneficial to improvement of discharge stabilities due to sufficient cooling of discharges at high pressures. A stable brush-shaped plasma jet was achieved by introducing a flowing argon through the narrow slit with two cylindrical electrodes set opposite at its ends and powered by a DC supply11. Here, a transverse electric field was applied on the flowing gas with its flow direction perpendicular to the field. The spatially resolved optical emission spectroscopy (OES) of the plasma revealed that the negative glow, positive column and anode glow exist across the gas gap. Li et al. employed the similar device as that used in Ref. 11 to generate a brush-shaped plasma jet12, where the same spatial discharge structure was observed. Another electrode configuration was reported by Lu et al.13. A high-voltage pulsed DC power supply was connected to the blades that serve as the anode and the treated objects were connected to the ground, where a longitudinal electric field was applied on the flowing gas. The negative glow, Faraday dark space and positive column in the brush-shaped plasma are clearly visible to the naked eye. Additionally, a DC-driven plasma source equipped with hollow-needle-plate electrodes was developed by Li et al.14, where the gas flows along the cavity of the hollow needle and rushes to the plate. This plasma presents us a diffuse plume in cup shape in the discharge space. As mentioned above, in order to obtain a stable plasma, the working gas was allowed to flow through the discharge space with either a transverse or a longitudinal electric field applied on. As for the longitudinal electric field, the working gas flows from one electrode to the other, where the needle-plate (or rod-plate) electrodes are usually used. In this work, a plasma jet is generated by using a longitudinal electric field that is generated between two rod electrodes. This plasma shows quite different physical structure in comparison with those observed in the traditional DC glow discharge.

Results

To demonstrate the unique DC-driven plasma jet, we used the atmospheric-pressure plasma device sketched in Fig. 1. The voltage-current (V-I) characteristic of the DC discharge was examined at the gas flow 2 L/min, with the relationship between the discharge voltage and current shown in Fig. 2. It is found that only subnormal (zone I) and normal (zone II) glow discharge modes appear in the examined current scope of 9–33 mA, where the Townsend mode is not observed. In the subnormal glow discharge (9–18 mA), the V–I curve assumes a negative differential resistance and the discharge voltage across the gas gap drops with the increasing current. When the discharge is transformed to the normal one (18–33 mA), the V–I curve becomes essentially flat and the discharge voltage approximately maintains a constant with the increasing current.

Figure 1
figure 1

The experimental setup.

Full size image
Figure 2
figure 2

Voltage-current characteristic of the plasma source with 2 L/min argon as working gas at atmospheric pressure.

Full size image

Figure 3(a) shows the experimental apparatus without discharge. When the apparatus works at an operating point with the discharge voltage of 234 V and current of 22 mA, the discharge operates in the normal glow mode and a stable plasma jet is formed between the two electrodes, as shown in Fig. 3(b). The plasma photograph was taken by using a digital camera (Nikon D5200) with the exposure time of 20 ms. The plasma appears to eject out of the quartz tube from the end face of the anode and run towards the cathode and gradually becomes thick from the top to the bottom, looking like an inverted torch. The diameter of the plasma jet is approximately linearly increased from 1 mm close to the anode to about 3 mm around the cathode, exhibiting an increasing cross-section. A closer observation indicates that parts of the plasma run out of the discharge space between the electrodes along the gas flow direction and surround the cathode end face. Figure 3(c) shows the isointensity map of the plasma image out of the quartz tube (with x varied from 1 to 7 mm). It is found that the luminance of the plasma is weakened along the radial direction.

Figure 3
figure 3

Photographs of (a) the experimental apparatus and (b) the plasma jet generated between two rod electrodes with the electrode spacing of 6 mm (with the exposure time of 20 ms).(c) shows the isointensity map of the plasma image out of the quartz tube.

Full size image

To investigate reactive species in the plasma, the OES was examined by using a moderate-resolution spectrometer with its fiber arranged at the specific spot (x = 3 mm and y = 0 mm) in accordance with the axe labeled in Fig. 3. Typical emission spectra with the wavelength range from 200 to 1000 nm are depicted in Fig. 4. In the ultraviolet and visible region, between 200 and 450 nm, the spectra are dominated by N2(C3Πu → B3Πg) emissions. Additionally, the OH emission lines, resulted from OH(A2Σ+ → X2Π) transition, are also visible. In the visible and infrared region between 690 and 950 nm, we can easily observe various excited argon atomic lines (e.g., λ = 763.7, 772.7 and 811.7 nm) from the 4 p → 4 s transitions. Meanwhile, a group of oxygen atomic lines around 777.8 nm, which correspond to O(I) triplet state transitions (O(5P) → O(5S0)), are also observed in close wavelength to one another, but cannot be distinguished due to the resolution limit of the spectrometer.

Figure 4
figure 4

The emission spectrum of the plasma jet examined at the spot with x = 3 mm and y = 0 mm.

Full size image

Due to frequent collisions among heavy particles at atmospheric pressure, the rotational temperature (Trot) of the OH radical has a value which is generally close to the gas temperature15,16. This temperature can be measured by using the OH(A2Σ+ → X2Π, Δv = 0) band transition from 306 to 312 nm. This band is easily isolated under various experimental conditions to give useful information. The vibrational temperature (Tvib) provides the information of the energy transfer between electrons and heavy particles, especially molecules17. It was obtained by the typical Boltzmann plot method of the nitrogen second positive system N2(C3Πu → B3Πg, Δv = −2) in the range 370–382 nm18. Based on the emission spectrum measured by using the high-resolution spectrometer with the fiber probe arranged at the same spot (x = 3 mm and y = 0 mm), the rotational temperature and vibrational temperature of the plasma were obtained through the analysis of OH and N2 molecular emission spectrum, respectively. As seen from Fig. 5, the simulated spectra at Trot = 810 ± 32 K and Tvib = 2200 ± 45 K give a good fit to the experimental spectra, indicating that the vibrational temperature is higher than the rotational temperature.

Figure 5
figure 5

The comparison of experimental and simulated spectra of (a) OH(A2Σ+ → X2Π, Δv = 0) transition for Trot measurement and (b) N2(C3Πu → B3Πg, Δv = −2) transition for Tvib measurement.

Full size image

The spatially resolved OES of reactive species such as OH (309.7 nm), N2 (337.1 nm) and Ar (763.7 nm) was examined along the x axis or axial direction at y = 0 mm (from the anode to the cathode labeled in Fig. 3), with the result shown in Fig. 6(a), where the discharge current is 22 mA. It is found that the OH and Ar assume the similar spatial distribution of emission profile. Their OES intensity first increases along the x axis (0–1 mm) and then approximately maintains unvaried (1–4 mm). After a slight drop (4–5 mm), the intensity rises largely, reaches its maximum around x = 6 mm, afterwards gradually declines and tends to be zero near the cathode (6–7 mm). This spatial distribution feature of OES indirectly reveals that the negative glow (around x = 6 mm), Faraday dark space (around x = 5 mm) and positive column (1–4 mm) are distributed from the cathode to the anode. The OES intensity of N2 exhibits a quite different spatial distribution, in comparison with that of OH and Ar. Its intensity maintains very low and nearly unvaried from x = 0 to 3 mm. After a slow increase (3–5 mm), the intensity rises sharply (5–6 mm), approaches its maximum around 6 mm and drops rapidly near the cathode (6–7 mm).

Figure 6
figure 6

Emission profiles of the radical OH (309.7 nm), excited N2 (337.1 nm) and excited argon (763.7 nm) along x axis with a longitudinal (a) and transverse (b) electric field applied on the flowing gas, respectively.

Full size image

Figure 6(b) shows the OES spatial distribution of a stable brush-shaped plasma jet generated in an extended DC discharge by applying a transverse electric field on the flowing argon, which was presented in our previous work11. This DC discharge was supported by a DBD one and operated in a normal glow mode at discharge current 22 mA. The OES of the plasma was examined along the x axis (from the cathode to the anode) with the distance 2 mm away from the exit. The spatially resolved OES showed that all of these reactive species presented a near-symmetrical distribution of emission profile, with the greatest emission intensity at the edges.

The spatially resolved OES of these reactive species was also examined along the y axis with x = 3 and 6 mm labeled in Fig. 3. Figure 7 shows that the emission intensity of all the reactive species presents a symmetric distribution with its maximum at the center. Along the y direction at x = 3 mm, the light emitted from the excited argon exceeds that from other species and the N2 exhibits the weakest emission intensity, as shown in Fig. 7 (a). With x = 6 mm, as seen from Fig. 7 (b), the light emitted from excited N2 is relatively increased considerably and overwhelms both other species of light along the y axis. The OH emission intensity becomes the weakest one.

Figure 7
figure 7

Emission profiles of the radical OH (309.7 nm), excited N2 (337.1 nm) and excited argon (763.7 nm) along the y axis with x fixed at (a) 3 mm and (b) 6 mm, respectively.

Full size image

The spatial distribution of rotational and vibrational temperature was measured through the spectral fitting method along the x axis at y = 0 mm, with the results depicted in Fig. 8. The errors in the measured rotational and vibrational temperature are less than ±40 and ±50 K, respectively. As seen from Fig. 8 (a), the highest rotational temperature (1050 K) is located in the negative glow region and there is a decrease of temperature close to the Faraday dark space. After the Faraday dark space, this temperature rises gradually and approaches about 810 K in the positive column. Approaching the anode, the temperature decreases to about 630 K. Figure 8 (b) shows a distinct increase of vibrational temperature from the negative glow region to the Faraday dark space where the temperature arrives at its maximum of 2450 K. This temperature reduces after the Faraday dark space and reaches about 2200 K in the positive column. The measured rotational and vibrational temperature profiles are in well agreement with those reported by Staack et al.19.

Figure 8
figure 8

The spatial distributions of (a) rotational temperature and (b) vibrational temperature along the x axis at y = 0 mm.

Full size image

Discussion

In the traditional DC normal glow discharge in a still argon at atmospheric pressure, the contraction of the positive column, which is attributed to the recombination process, allows the gas discharge to occur only in the discharge space between two opposite plane electrodes6,20. The cross-section of the positive column has been changed little along the axial direction (from the anode to the cathode)10. Comparison with the torch-shaped plasma jet in our case (see Fig. 3(b)) indicates that the gas flow plays a key role in variation of the plasma structure and contributes much to enlarging the plasma volume along the radial direction especially in the vicinity of the cathode. In the work performed by Li et al.14, a cup-shaped plasma could be formed between the hollow-needle anode and the plate cathode. The backflow gas and the large-area plate cathode are responsible for the large-volume diffuse plasma generated in the vicinity of the cathode. In our case, the circular hollow metal base avoids the gas backflow and gas dispersion to a larger open space around the discharge region. This promotes the formation of a thin gas flow channel around the column of the discharge region, leading to the generation of plasma with a high-power volume density in a limited space. This feature is crucial in applications such as surface modification, where a high-power-density plasma is required.

As we know, reactive species, such as OH and O, are the indispensable components and main enablers for industrial and medical plasma treatment applications. The OES of the torch-shaped plasma jet indicates that many reactive species, including OH, O, excited N2 and argon metastables, are produced in the glow discharge, as shown in Fig. 4. All these reactive species are desirable in the applications such as surface modification and biological decontamination. Since the plasma jet is generated in the ambient air, the presence of OH, O and excited N2 spectra is attributed to the interaction of the argon plasma with the ambient air that contains nitrogen, oxygen, water vapor and other components, which will be addressed in detail in the following section.

Examination of the rotational temperature and vibrational temperature of the plasma in Fig. 5 shows that the plasma jet is under nonequilibrium condition. This feature is favorable for enhanced plasma chemistry. The electronic excitation temperature Texc of plasma is determined according to the double-line method21:

where two close spectral lines (λ1 = 763.7 nm, 2P6 → 1S5 and λ2 = 772.7 nm, 2P2 → 1S3) are used; excitation potentials of the spectral lines ξ1 = 13.17 eV and ξ2 = 13.328 eV; the coefficient α is 5040 for ξ in eV; the weighting factor g1 = 5 and g2 = 3; transition probability of the two spectral lines A1 = 24.5 and A2 = 11.7; and i is the relative intensity of spectral line with i1 ≈ 45800 and i2 ≈ 1710022, The electronic excitation temperature is estimated to be 0.57 ± 0.02 eV (6586 ± 240 K) and far beyond the vibrational temperature. This also indicates the nonequilibrium characteristic of the plasma jet.

However, the vibrational temperature determined through analysis of N2 emission spectra as a result of electron impact excitation is usually measured to be about 3000 K23 and much higher than the temperature examined in our case. This suggests that the excited species OH(A2Σ+), O(5P) and N2(C3Πu) are partly generated by energy transfer from argon metastables, in addition to the electron excitation. As is well known, the energy carried by the argon metastables (11.5–13.5 eV)24, which are produced effectively in our case indicated by the spectrum shown in Fig. 4, is slightly higher than that of energy level of N2(C3Πu) (~11.1 eV) from the ground state25. The argon metastables can easily transfer energy to the ground state nitrogen molecules to generate abundant excited state N2(C3Πu) molecules. Besides, the energy required to dissociate H2C from the ambient air into OH and H and subsequently excite the ground state OH to the excited state OH(A2Σ+) is only 5.1 and 4.2 eV respectively24. Sufficient excited state OH(A2Σ+) molecules can be produced by energy transfer from argon metastabels. Thus, the emission spectra arising from the excited OH and N2 are distinctively presented. The energy threshold is 15.9 eV for the production of O(5P) from O2 and is beyond the energy level of the argon metastables25. Therefore, the direct generation of excited O from O2 through energy transfer with excited argon atoms is ignorable in our case. However, the energy for dissociation of an oxygen molecule to oxygen atoms is 5.58 eV and that required to subsequently produce O(5P) from ground state O is about 10.7 eV25,26. Since the two potential levels are in the range of the energy carried by argon metastables, the excited O can be produced from their ground state atoms through energy transfer processes with excited Ar. As a result, the oxygen emission spectra are observed.

The spatially resolved OES of the plasma in Fig. 6(a) presents the spatial emission spectrum distribution of reactive species. As discussed above, the excited OH and N2 are produced from water vapor and nitrogen in the ambient air through complex reaction channels. The corresponding reaction coefficients are largely dependent on the reaction conditions, such as reactant concentration and energy required for reaction onset. The distinct spatial distribution of OH and N2 emission spectra is probable due to the largely different volume percent between water vapor and nitrogen in ambient air and different energy required to pump them from the ground state to an excited one16. In our experiment, the argon ejects into the ambient air from the end face of the anode with a certain velocity and mixes with the ambient air. The un-fully mixing partly contributes to the low and slowly increasing intensity of N2 emission spectra in the positive column. In addition, the gas flow brings parts of the excited N2 downward and few are left upward in the positive column. This contributes much to assembling most of the excited N2 in the vicinity of the cathode.

In our previous work, as shown in Fig. 6(b), a near-symmetrical distribution of emission profile of reactive species (radical OH, excited N2 and excited argon) was presented in front of us, with a flat area located in the middle and two peaks situated at the edges. It is considered that the flat area is correlated with the positive column of the DC glow discharge, while the two peaks at the edges is attributed to the negative and anode glow, where there exist many energetic electrons. In this work, the near-symmetrical distribution of emission profile cannot be found, because the peak of emission intensity appears only near the cathode. The absence of the emission intensity peak near the anode suggests that the anode glow does not exist.

Since the gas flow is fixed at 2 L/min and the inner diameter of quartz tube is 1.7 mm, the gas flow velocity near the anode is estimated to be about 1.7 × 103 cm/s. The electric field in the positive column can be expressed as E = (Vdis − Vn)/D, with the value of 172 V/cm. Here, Vdis is the sustaining voltage (234 V) between the electrodes, Vn is the cathode fall (~165 V) for the iron electrode6 and D is the length of the positive column (~4 mm). The electron drift velocity is given by ve = μeE. With the electron mobility μe of 4.3 × 102 cm2/(V·s)6, the electron drift velocity is approximately equal to 0.7 × 105 cm/s and far beyond the gas flow velocity. This means the gas flow has practically no direct effect on the motion of electrons. As far as the ion motion is concerned, the ion drift velocity vi = μiE≈ 0.3 × 103 cm/s, where only one sort of ions Ar+ is considered for simplicity and the ion mobility μi = 1.9 cm2/(V·s)27. It is found that the gas flow velocity can be comparable with the ion drift one and approximately 6 times higher than the latter. Thus, the gas flow has an appreciable effect on the motion of ions. In our case, both the gas and the ions flow from the anode to the cathode and they drift in the same direction, while the electrons travel in the opposite direction. Due to the higher gas flow, acceleration of ions in the positive column is greatly enhanced, so that overmany ions flux through the cross section of positive column per unit time and drift along the electric field. Subsequently, Ion vacancies are formed in the local discharge space. If these ion vacancies are not occupied in time by new ions, parts of electrons will reverse their drift direction and run after the drifting ions by the electrostatic force due to the regime of ambipolar diffusion6,28. Under such conditions, many extra ions in the vicinity of the anode are brought into the positive column by the gas flow. The electrons thus follow the ions there. As a result, the electron density near the anode is substantially reduced and the anode glow is absent, showing us an asymmetrical distribution of emission profile.

The emission profile of the reactive species shown in Fig. 7 reveals that the occurrence of chemical reactions are concentrated in the center part of the plasma and most of the reactive species are produced there. When these reactive species are deexcited, they emit a brighter light, as illustrated in Fig. 3(c).

The spatially resolved rotational and vibrational temperature measurements in Fig. 8 show that the discharge is nonthermal throughout the discharge space. The peak and valley in the profile of rotational temperature are associated with heating of the gas by energetic electrons6. The rotational temperature peak in the negative glow region is attributed to the more energetic electrons due to the high electric field there. In the Faraday dark space, the electrons are less energetic with a low electric field, resulting in a falling of temperature. The decrease of rotational temperature close to the anode is probably due to the higher gas flow velocity which is beneficial to cooling the plasma. A correspondence between higher vibrational temperature and lower rotational temperature is observed probably due to the increased vibration-translation relaxation rates at higher gas temperature19,29.

Methods

A schematic of the experiment setup is shown in Fig. 1. The plasma source mainly comprises two parts. One is the anode unit and the other is the cathode unit. In the anode unit, a cylindrical metal rod (with the diameter of 1 mm) serves as the anode that is centered inside a quartz tube (with the inner diameter of 1.7 mm and outer diameter of 3.3 mm). The metal rod is connected to a high-voltage DC power supply through a 100 MΩ ballast resistor which is used to limit the discharge current. As for the cathode unit, a metal (iron) spring acts as the cathode and is embedded in the hollow cylindrical metal base (with the hollow diameter of 7 mm) that is connected to the ground. It should be noted that the metal spring is fixed tightly in the metal base and could hardly be deformed by the gas flow. The anode unit is set above the cathode one with the electrode spacing of 6 mm. Argon (with the purity of 99.999%), which is controlled by a mass flow controller (fixed at 2 L/min), enters by inlet port, flows through the circular region between the quartz tube and the anode and rushes out into the discharge space, where argon and ambient air are mixed. The volume fraction of water vapor in ambient air is estimated to be 1.9%, based on the temperature of 25°C and the relative humidity of 60% recorded by a humidity/temperature transmitter. With a proper voltage applied to the two electrodes, a stable plasma jet, indicated by the red pattern, can be formed between them. In our experiment, the gas mixture flows from the anode to the cathode and the flow has the same direction with the electric field. After travelling across the discharge space, the gas mixture will flow along the original direction and into the circular hollow space of the metal base, rather than flow back in the case where needle-plate (or rod-plate) electrodes are used. The flow behavior of argon is indicated by the light blue bar with an arrow downstream. The plasma is imaged with a digital camera (Nikon D5200). The emission spectra from the plasma are recorded by using two different resolution spectrometers. For the large emission spectrum range (200–1000 nm), an Avantes spectrometer (AvaSpec-ULS2048-USB2) with a moderate resolution of 0.8 nm is used to identify various reactive species. In order to obtain the rotational and vibrational temperature of the plasma, another Avantes spectrometer (AvaSpec-ULS2048-3-usb2) with a high resolution of 0.1 nm is used. Both their fiber probes are placed at 1.0 cm away and perpendicular to the plasma. The radius of observation spot of fiber probe is 2.3 mm, with the fiber diameter of 200 μm and the fiber numerical aperture of 0.22.