Introduction

The hollow cathode (HC) discharge was discovered by Bartels and Paschen more than 100 years ago [1]. The discharge is sustained by electrons undergoing a pendulum motion inside the HC [2,3,4,5,6,7]. Hollow cathode discharges are employed in many applications, e.g., in atomic spectroscopy [1, 8,9,10], mass spectrometry [11], and laser technology [12], as UV generators [13, 14] and propulsion thrusters [15, 16], for the generation of neutral and ion beams [17, 18], surface processing [19], thin film deposition [20,21,22,23,24,25,26,27], and in plasma chemical investigations [28,29,30,31,32,33,34].

Formation of nitric oxide (NO) has attracted much attention in previous years. Nitric oxide (NO) is a colourless toxic gas which forms by oxidation of nitrogen. It has rather few industrial applications. NO plays a role in combustion and is a flue gas generated, e.g., by automotive petrol engines and fossil fuel power plants [35, 36]. In general, production of nitric oxide during combustion is easier to achieve compared to its removal from the exhaust gas [37,38,39]. Nitric oxide is naturally produced by lightning in thunderstorms and in the upper mesosphere and lower thermosphere (65–140 km) by precipitating energetic electrons and protons during space weather events [40].

Nitric oxide is a small molecule which has wide range of physiological functions in living organisms. In plants, nitric oxide regulates plant metabolism and is involved in many physiological processes, in particular, germination, flowering, or leaf senescence [41, 42]. In mammals, NO is produced in cavernosal nerves and endothelial cells by nitric oxide synthase (NOS) enzymes. NO shows beneficial effect on the healing of skin wounds [43], regulates several physiological procesesses, e.g., blood pressure and hormone release, and plays a key role as a vasodilator of blood vessels and in the physiology of penile erection [44,45,46]. In 1998, the Nobel Prize in Physiology or Medicine 1998 was awarded jointly to Robert F. Furchgott, Louis J. Ignarro and Ferid Murad "for their discoveries concerning nitric oxide as a signalling molecule in the cardiovascular system" [47, 48].

NO formation via plasma chemical reactions has been investigated with different discharges, e.g., glow [49], streamer [50], arc [51], dielectric barrier [52], and microwave discharges [53,54,55,56]. In the present communication we utilize a pulsed hollow cathode (PHC) discharge for the investigation of plasma chemical reactions in a dilute Ar+N\(_2\)+O\(_2\) gas environment. We expect that different nitrogen oxides, in particular, nitric oxide (NO) and nitrogen dioxide (NO\(_2\)) will be generated. The use of a hollow cathode has several advantages compared to other plasma devices. Firstly, the plasma is confined by the walls of the hollow cathode which should give rise to large radical densities without the need of gas heating. Secondly, the properly chosen wall material could serve as a catalyst to further promote the desired chemical reactions [57]. The PHC is operated with a pulse length of 100 \(\mu\)s and a repetition frequency of 5 kHz (duty cycle 50 %). Plasma characterisation and, in particular, plasma chemical processes are studied with the help of energy-resolved mass spectrometry, by optical emission spectroscopy, and by Fourier transform infrared (FTIR) spectroscopy of the exhaust gas.

Experiment

The experimental set-up has been described before [26, 58, 59]. Very briefly, the set-up consists of a cylindrical hollow cathode (HC) inside a vacuum chamber (Fig. 1). The HC is made from copper nickel (Cu50Ni50) alloy. Argon, oxygen, and nitrogen gas are introduced into the vacuum chamber through the hollow cathode. The Ar gas flow rate is typically set to 200 sccm; nitrogen and oxygen gas flow rates are varied up to 100 sccm. The gas pressure directly at the nozzle outlet and inside the HC is significantly larger and in the range of a few 100 Pa. Under these conditions, the expected gas flow speed is sonic at the nozzle’s outlet and becomes supersonic inside the vacuum chamber [22, 26]. The large flow velocity could cause a periodic structure with a barrel shock created by a system of shock waves in the plasma inside the reactor chamber, as observed by Tichy et al. [6].

Fig. 1
figure 1

a Experimental set-up with vacuum chamber showing arrangement of hollow cathode (HC), anode, Langmuir probe, EQP analyser, optical window, and optical spectrograph (schematic). b Electronic arrangement showing cathode power supply, electronic switch, 100 \(\Omega\) resistor, pulse generator (5 kHz), and anode power supply

Fig. 2
figure 2

Temporal evolution of a discharge voltage and b discharge current of a PHC discharge with a Cu50Ni50 nozzle. Pulse length 100 \(\mu\)s, repetition frequency 5 kHz, anode voltage +30 V, mean discharge current 0.25 A. Argon gas flow rate 200 sccm, nitrogen gas flow rate 10 sccm, oxygen gas flow rate 10 sccm, Ar gas pressure \(p=2.1\) Pa

The hollow cathode is connected via a 100 \(\Omega\) resistor to a home-built power switch which is powered by a direct current (DC) power supply. The repetition frequency f and the pulse length T of the pulsed power switch are set to \(f=5\) kHz and \(T_{\textrm{on}}=100\,\mu\)s. The pulsed hollow cathode (PHC) discharge is operated with a mean discharge current of 0.25 A. During operation, the HC heats up to a temperature around 1,000 K as indicated by a red color. An optional ring anode with an inner diameter of 63 mm is mounted at a distance of 100 mm from the hollow cathode [58,59,60].

Energy-resolved mass spectrometry is performed with a commercial Hiden EQP 1000 mass/energy analyzer [61, 62]. The instrument is mounted opposite to the hollow cathode and the anode [58,59,60].

Optical emission spectroscopy (OES) is carried out with a Shamrock SR500D spectrometer equipped with an iCCD detector. The spectrometer is equipped with three gratings having 600 lines/mm, 1800 lines/mm, and 2400 lines/mm and blaze wavelengths of 500 nm, 500 nm, and 300 nm, respectively. An optical fibre connected to the entrance slit of the spectrometer is installed outside the vacuum chamber. It views the open end of the HC at an angle of 38\(^{\textrm{o}}\) with respect to the hollow cathode’s axis through a quartz window with a cut-off wavelength below 200 nm.

Ex-situ Fourier-transform infrared (FTIR) spectoscopy is employed to analyse the exhaust gas from the plasma chamber. Exhaust gas constituents are trapped in a liquid nitrogen-cooled glass container which is inserted into the exhaust line of the pumping system between the turbomolecular pump and the roughing pump. After removal from the exhaust line, the cold gas container is connected to a glass cell with ZnSe windows and optical path of 28 cm. The cell is inserted into the FTIR spectrometer where the sample warms up to room temperature. A high-resolution infrared Fourier-transform spectrometer (Bruker IFS 125HR, Bruker GmBH, Karlsruhe, Germany) is operated in the mid-infrared region between 600 cm\(^{-1}\) and 6000 cm\(^{-1}\) using a KBr beamsplitter and HgCdTe liquid-N\(_2\)-cooled detector. The measured spectra are averaged over 10 scans [63, 64]. The spectra are apodized with a Boxcar apodization function and analysed using the spectr library to obtain partial pressures of the detected components [65].

Results

Discharge Characteristics

The discharge voltage required to maintain a mean discharge current of 0.25 A depends on the employed gas mixture (Fig. 2). Pure Ar requires the smallest discharge voltage of about −425 V. Larger discharge voltages of about −470 V and −490 V are needed for the Ar+O\(_2\) and Ar+O\(_2\)+N\(_2\), respectively. The largest discharge voltage of –540 V is required to sustain the Ar+N\(_2\) gas mixture.

Except for some overshoots that occur right at the beginning of the pulse, all discharge currents increase monotonically with time up to the end of the pulse. Some differences between the gas mixtures are noted. In the case of pure Ar, the discharge current increases smoothly, while in the case of the gas mixtures Ar+O\(_2\), Ar+N\(_2\) and Ar+O\(_2\)+N\(_2\), a kink occurs after about 45 \(\mu\)s, followed by a gradual leveling off. No explanation for this behaviour has been found yet.

Fig. 3
figure 3

Ion mass spectra of a PHC discharge with a Cu50Ni50 nozzle operated in a Ar, b Ar+N\(_2\), c Ar+O\(_2\), and d Ar+N\(_2\)+O\(_2\) gas mixtures. Pulse length 100 \(\mu\)s, repetition frequency 5 kHz, anode voltage +30 V, mean discharge current 0.25 A. Argon gas flow rate 200 sccm, nitrogen gas flow rate 10 sccm, oxygen gas flow rate 10 sccm, Ar gas pressure \(p=2.1\) Pa. Detected ion energy 27 eV. Vertical dashed lines indicate the \(m/z=30\) position

Fig. 4
figure 4

Ion energy distribution of N\(_2^+\) (\(\circ )\), NO\(^+\) (\(\triangle\)), O\(_2^+\) (\(\diamond )\), Ar\(^+\) (\(\triangledown\)), and NO\(_2^+\) ions for a PHC discharge with a positively biased anode (+30 V) in an Ar+O\(_2\)+N\(_2\) gas mixture. Ar gas flow rate 200 sccm, O\(_2\) gas flow rate 10 sccm, N\(_2\) gas flow rate 10 sccm, Ar gas pressure 2.2 Pa, discharge current 0.25 A

Ion Mass and Ion Energy Distribution

Typical ion mass spectra of a PHC discharge operated in different gas mixtures are displayed in Fig. 3. The mass spectrum from pure Ar is dominated by singly-charged Ar\(^+\) (\(m/z=40\)), Ar\(^{2+}\) (\(m/z=20\)), Ni\(^+\) (\(m/z=58\) and 60) and Cu\(^+\) (\(m/z=63\) and 65) ions, where m and z are ion mass and charge number, respectively. The spectrum also contains a mass peak at \(m/z =18\) which is attributed to H\(_2\)O\(^+\) impurity ions. The mass spectrum from the Ar+N\(_2\) or Ar+O\(_2\) gas mixtures additionally display strong mass peaks of N\(_2^+\) (\(m/z=28\)) or O\(_2^+\) (\(m/z=32\)), respectively, ions. Also present are peaks at \(m/z=14\) and \(m/z=16\) of N\(^+\) and O\(^+\) ions, respectively. The most dominant mass peak of the Ar+N\(_2\)+O\(_2\) gas mixture at \(m/z=30\) is attributed to the formation of NO\(^+\) ions. This mass peak is virtually absent in the other mass spectra. In addition, formation of NO\(_2^+\) (\(m/z=46\)) is noted. Compared to the ion mass spectra from the Ar+N\(_2\) or Ar+O\(_2\) gas mixtures, the intensities of N\(_2^+\) or O\(_2^+\) ions, respectively, are considerably reduced. This is a clear indication that NO\(^+\) ions (and NO molecules) are generated through plasma chemical reactions of nitrogen and oxygen species.

Figure 4 compares ion energy distributions of N\(_2^+\) (\(m/z=14\)), O\(_2^+\) (\(m/z=32\)), and Ar\(^+\) (\(m/z=40\)) ions obtained with the Ar+N\(_2\)+O\(_2\) gas mixture. Also shown are results for NO\(^+\) (\(m/z=30\)) and NO\(_2^+\) (\(m/z=46\)) ions which originate from plasma chemical reactions. A positively biased anode (+30 V) to stabilize the discharge and the enhance the kinetic energy of plasma ions is employed during these measurements. The intensity maximum (peak) of the ion energy distributions is shifted to a kinetic energy close to \(E=e_0 V_p\), where \(e_0\) is the elementary charge and \(V_p\) is the plasma potential which is controlled by the anode voltage [58, 59].

The measured energy distributions show a complicated structure with typically two maxima. The energetic position of the high-energy maximum corresponds to the plasma potential close to the anode. The measured kinetic energy where the high-energy maximum occurs is somewhat smaller than expected from the applied anode voltage. It is a typical phenomenon for an oxygen-containing disharge and, e.g., not observed for a HC discharge in pure argon. The Ar\(^+\) ions display a pronounced low-energy maximum which is caused by resonant charge exchange reactions and thus does not represent the plasma potential in the vicinity of the anode [58].

Fig. 5
figure 5

Energy-integrated intensity of NO\(^+\) (\(\bullet\)), NO\(_2^+\) (\(\blacktriangle\)), O\(_2^+\) (\(\diamond\)), Ar\(^+\) (\(\triangledown\)), and N\(_2^+\) (\(\square\)) ions versus N\(_2\) gas flow rate. Ar gas flow rate 200 sccm, Ar gas pressure 2.1 Pa, O\(_2\) gas flow rate 10 sccm

Fig. 6
figure 6

Energy-integrated intensity of NO\(^+\) (\(\blacktriangle\)), NO\(_2^+\) (\(\blacktriangledown\)), N\(_2^+\) (\(\circ\)), Ar\(^+\) (\(\diamond\)), and O\(_2^+\) (\(\square\)) ions versus O\(_2\) gas flow rate. Ar gas flow rate 200 sccm, Ar gas pressure 2.1 Pa, N\(_2\) gas flow rate 10 sccm

Fig. 7
figure 7

Optical emission spectrum of a PHC discharge with a Cu50Ni50 nozzle operated in a Ar, b Ar+N\(_2\), and c Ar+N\(_2\)+O\(_2\) gas mixtures. Wavelengths of (\(\nu '=0,\nu ''=0-4\)), (\(\nu '=1,\nu ''=1-5\)), and (\(\nu '=2,\nu ''=2-7\)) transitions of the NO \(\gamma\)-bands are indicated. Optical grating with 2,400 lines/mm and a blaze wavelength of 300 nm. Pulse length 100 \(\mu\)s, repetition frequency 5 kHz, mean discharge current 0.25 A. Argon gas flow rate 200 sccm, nitrogen gas flow rate 10 sccm, oxygen gas flow rate 10 sccm, Ar gas pressure \(p=2.1\) Pa

Energy-integrated ion intensities are displayed in Figs. 5 and 6 as function of N\(_2\) and O\(_2\), respectively, gas flow rates. The intensity of N\(_2^+\) ions shows an approximately linear increase with increasing N\(_2\) gas flow rate (Fig. 5). The intensity of NO\(^+\) ions monotonically increases with increasing N\(_2\) gas flow rate. The intensity of NO\(_2^+\) ions also increases with increasing N\(_2\) gas flow rate, however, reaches a maximum at a gas flow rate of \(\approx 2\) sccm and continues with a pronounced decline. The O\(_2^+\) intensity shows a monotonic decrease while the Ar\(^+\) intensity is little influenced by the N\(_2\) gas flow rate. It indicates that O\(_2\) is consumed by the formation of NO\(^+\) ions.

As a function of the O\(_2\) gas flow rate (Fig. 6), the NO\(^+\) ion intensity shows an initial increase, saturates at a gas flow rate of \(\approx 15\) sccm, and remains essentially constant beyond this point. N\(_2^+\) and Ar\(^+\) ion intensities gradually decrease with increasing O\(_2\) flow rate. It may reflect the changing plasma and surface conditions due to the admittance of O\(_2\) through the cathode.

Optical Emission Spectra

Optical emission spectra in the wavelength range 220–280 nm are displayed in Fig. 7. It is the region where we expect the \(\gamma\)-bands (A \(^2\Sigma\) \(\rightarrow\) X \(^2\Pi _r\)) of nitric oxide [66,67,68,69]. The pure Ar spectrum shows several prominent neutral copper (Cu I) lines at 222.57 nm, 244.16 nm, 249.21 nm, 261.84 nm, and 276.64 nm [70, 71]. Neutral nickel (Ni I) lines are observed at 230.08 nm, 231.23/231.37/231.40 nm, 232.00/232.14 nm, and 234.55 nm. Ionized copper (Cu II) lines are present at 224.70 nm, 227.63 nm, and 236.99 nm. Ionized nickel (Ni II) lines are observed at 228.77 nm, 237.54 nm, 239.45 nm, 241.61 nm, and 243.79 nm.

Similar conclusions can be drawn from the emission spectrum of the Ar+N\(_2\) gas mixture. By contrast, the emission spectrum of the Ar+N\(_2\)+O\(_2\) gas mixture is largely different. The spectrum displays several broad emission bands which are attributed to the NO \(\gamma\)-bands. The band heads of the relevant (\(\nu '\), \(\nu ''\)) transitions [72] are indicated in the lower part of Fig. 7c. The present result is a clear indication for the formation of NO molecules in a hollow cathode discharge. Optical emission spectra for different N\(_2\) gas flow rates measured in the spectral range 249–276 nm are displayed in Fig. 8. A linear increase of the integrated OES intensity with N\(_2\) gas flow rate is observed.

Fig. 8
figure 8

Optical emission spectrum of a PHC discharge with a Cu50Ni50 nozzle operated in an Ar+N\(_2\)+O\(_2\) gas mixture for different nitrogen gas flow rates. Wavelengths of (\(\nu '=0,\nu ''=0-4\)), (\(\nu '=1,\nu ''=1-5\)), and (\(\nu '=2,\nu ''=2-7\)) transitions of the NO \(\gamma\)-bands are indicated. Optical grating with 600 lines/mm and a blaze wavelength of 500 nm. Pulse length 100 \(\mu\)s, repetition frequency 5 kHz, mean discharge current 0.25 A. Argon gas flow rate 200 sccm, oxygen gas flow rate 50 sccm, Ar gas pressure \(p=2.1\) Pa

Samples taken from the exhaust line are examined by FTIR spectroscopy. Gas flow rates of 200 sccm, 10 sccm, and 10 sccm for Ar, N\(_2\), and O\(_2\), respectively, are employed and the discharge current is set to 0.25 A. Nitric oxide has a melting point of 109 K [73] and, hence, is, readily captured by the liquid-nitrogen-cooled surface. The analysed spectrum is shown in Fig. 9. The total volume of the sampled gas (at standard conditions) after 60 min of discharge operation is about 80 cm\(^3\). The sample contains a large amount of NO, NO\(_2\), and a small fraction of N\(_2\)O. The relative composition of the observed nitrogen oxides are shown in table 1. In addition, a significant amount of water (H\(_2\)O, about 20 cm\(^3\)) and a rather small amount of CO\(_2\) are present whose origins are not known.

Fig. 9
figure 9

FTIR of a PHC discharge with a Cu50Ni50 nozzle. a Full range and b expanded range showing the NO\(_2\) band at 2907 cm\(^{-1}\). Sampling time 60 min. Pulse length 100 \(\mu\)s, repetition frequency 5 kHz, anode voltage +30 V, mean discharge current 0.25 A. Argon gas flow rate 200 sccm, nitrogen gas flow rate 10 sccm, oxygen gas flow rate 10 sccm, gas pressure \(p=2.3\) Pa

Discussion

The formation of NO molecules in a N\(_2\) + O\(_2\) gas mixture requires the dissociation of N\(_2\) and O\(_2\) molecules into N and O radicals, either by thermal processes at suffiently high gas temperatures, or, as is the case here, in a non-thermal plasma by electron impact [74],

$$\begin{aligned} \textrm{e}^- + \textrm{X}_2 \rightarrow \textrm{e}^- + \textrm{X} + \textrm{X} \,, \end{aligned}$$
(1)

where X stands for N or O. Once formed, N and O radicals can interact with O\(_2\) and N\(_2\), respectively, molecules via the so-called Zeldovich mechanism [74,75,76,77,78,79,80,81]

$$\begin{aligned} \textrm{N} + \textrm{O}_2 \rightarrow \textrm{NO} + \textrm{O} \end{aligned}$$
(2)

and

$$\begin{aligned} \textrm{O} + \textrm{N}_2 \rightarrow \textrm{NO} + \textrm{N} \,. \end{aligned}$$
(3)

The Zeldovich mechanism is one of the most important reactions for nitric oxide (NO) formation. At the gas temperatures of interest here, reaction 3 is more important since due to the about one order of magnitude smaller activation energy its rate coefficient is several orders of magnitude larger compared to reaction 2.

In the following we anticipate that the majority of the detected ions has formed via neutral-neutral reactions and subsequently are ionised by electron impact in the plasma. The observed ion intensities, hence, reflect the densities of the corresponding neutral species. The ion intensities shown in Fig. 5 indicate that NO formation is a function of N\(_2\) gas density and as such increases with the N\(_2\) gas flow rate. Simultaneously, the density of O\(_2\) decreases as corroborated by the decreasing O\(_2^+\) intensity. Evidently, much of the supplied oxygen is consumed.

Although only a fraction of N\(_2\) has been consumed so far, the NO\(^+\) ion intensity already reaches its saturation level at an O\(_2\) gas flow rate of about 15 sccm (Fig. 6). The saturation coincides with a pronounced increases of the NO\(_2^+\) density which is readily understood by the larger oxygen supply. It can be explained by a further oxidation of NO to NO\(_2\), e.g., via [82, 83]

$$\begin{aligned} \textrm{O} + \textrm{NO} + \textrm{M} \rightarrow \textrm{NO}_2 + \textrm{M} \,, \end{aligned}$$
(4)

where M = N\(_2\), O\(_2\), or Ar. A different behaviour is noted as a function of N\(_2\) gas flow rate where the NO\(_2^+\) intensity reaches a maximum at about 2 sccm and then declines (Fig. 5). With increasing N\(_2\) density the number of N radicals increases and back-reactions like

$$\begin{aligned} \textrm{N} + \textrm{NO}_2 \rightarrow \textrm{NO} + \textrm{NO} \,. \end{aligned}$$
(5)

may play a significant role [82, 84].

Table 1 Fraction of detected nitrogen oxide species (%)

Conclusions

Plasma-chemical reactions in a pulsed hollow cathode discharge were investigated. A gas mixture of Ar+N\(_2\)+O\(_2\) was used. The formation of nitrogen oxides, especially NO and NO\(_2\), was observed. NO is the dominant molecular species formed in the discharge. Other nitrogen oxides such as N\(_2\)O and NO\(_2\) are produced in smaller amounts. The proportion of other nitrogen oxides depends on the N\(_2\) and O\(_2\) gas flow rates and the N\(_2\) to O\(_2\) ratio. The total amount of nitrogen oxides, mostly NO and NO\(_2\), captured in the exhaust line with a liquid nitrogen-cooled glass container is about 60 cm\(^3\). The role played by the employed CuNi nozzle is not investigated and with respect to eventual catalytic properties remains as a future issue.