P. Scheffler¹, A. Bräuning-Demian², K.-H. Gericke¹, C. Geßner¹, C. Penache², H. Schmidt-Böcking², L. Spielberger²

Large-area direct current (DC) discharges are generated using micro-structured electrode (MSE) 2D-arrays with electrode distances between 50 and 250 mm. The arrays are manufactured by means of modern micromachining technologies. A homogenous diffuse light emission, which covers the whole active array surface with a typical thickness of approximately 0.5-1.5 mm, is observed. The discharge can be operated over a wide pressure range with sustaining voltages of only a few 100 Volt DC and shows the characteristics of a normal glow discharge. A fundamental description of the discharge by its appearance and electrical characteristics is given. Possible important areas of application of the MSE 2D-arrays are pollution control, surface modification of materials, plasma chemistry and the use as light sources. In order to demonstrate the manifold capabilities of the arrays a series of measurements has been performed which show their applicability to pollution control. For these investigations an experimental setup was realized for the detection of nitrogen oxide (NO) via laser induced fluorescence (LIF) technique.

1. Introduction

Non-thermal plasma processing techniques are well established in a wide range of applications e.g. pollution control [1], surface modification of materials [2] or lighting technology. The applied weakly ionized plasmas are highly non-equilibrium, i.e. the electrons have large mean kinetic energies equivalent to several 10000 K whereas the gas remains at ambient temperature. Various high energy sources like laser radiation, electron, ion, and neutral beams, X-ray or g -radiation allow to produce high energy electrons and thus non-thermal plasmas. However, the technically most important method to generate a non-thermal plasma is by high electric field (electrical discharge). For that, direct current (DC), alternating current (AC), or radio frequency (RF) up to the GHz range can be applied. Since the basic electrical discharge creation process in gases proceeds via electron impact ionization, electrons must be accelerated by the applied field E over the mean free path l to gain a minimum kinetic energy

where E_ex is the threshold energy necessary for electronic excitation, dissociation, fragmentation or ionization of the gas. Therefore, operation of electrical discharge devices requires either a low gas pressure, i.e. large l, or a very strong electric field E. For example, if an electron should gain a mean kinetic energy of 10 eV at atmospheric pressure, where l is about 10^-5 cm, E must exceed 10⁶ V/cm, which is very difficult to achieve with conventional electrode arrangements. Consequently, vacuum conditions and/or a high voltage environment are usually necessary to generate an electrical discharge. Under these circumstances the application of plasma technology is often quite complex and expensive. For example, it is very difficult to establish a continuous process management, if vacuum conditions are required. The search for discharge devices, which operate without any vacuum restraints and at moderate voltages, is therefore a constant challenge in plasma science.

One promising approach are electrode arrays fabricated by means of modern micromachining technologies, which can provide distances on a micron-scale between the electrode elements. Here, even at moderate voltages, the electrostatic field can exceed the threshold value required to initiate the electrical breakdown process. The field between the electrodes easily exceeds 10⁵ V/cm and can approach 10⁶ V/cm and even higher values at the edges of the electrodes. In this regard it is possible to produce gas discharges up to atmospheric pressure by means of micro-structured electrode (MSE) arrays with very moderate voltages so that a wide area of plasma applications becomes feasible under reasonable conditions. For instance, MSE generated discharges in rare gases can be used to clean surfaces. If the arrays are manufactured on flexible substrates even any non-planar surface can be treated. Similarly, MSE arrays can be used for plasma deposition of thin films on surfaces. Other conceivable applications are as a spectral light source in the micron-scale for spectroscopy measurements or for pollution control. One important advantage regarding the applicability of MSE arrays is the low-cost manufacturing by standard methods.

Two types of MSE arrays are under investigation. They have conceptually different electrode geometries as shown schematically in figure 1. Depending on the size of the electrode geometry large-area as well as micro-scale discharges can be created. The device shown at the top, which we denote as a MSE 3D-array, is based on the hollow cathode effect. It consists of an insulator serving as a spacer, coated with metal layers on both sides, which form the cathode and the anode, respectively. The thickness of the insulator as well as that of the metal coatings is in the micrometer range. In this composite sheet a matrix of cylindrical channels with a diameter of 40-500 mm is manufactured by chemical etching on both sides. MSE 3D-arrays are operated with DC. A detailed description of the MSE 3D-arrays studied in our groups is given in [3].

The aim of this paper is to introduce the concept of the MSE 2D-arrays shown schematically at the bottom of figure 1. In order to demonstrate the capability of the 2D-arrays for technical applications a series of measurements has been performed, which show their applicability to pollution control. The results are discussed herein. Apart from that a description of the discharge by its appearance and electrical characteristics is given.

The original field of application of the described MSE 2D-arrays was in nuclear physics as a new kind of position-sensitive discharge amplifiers [4] to build new generation gaseous position-sensitive proportional counters and to replace the traditional multi-wire drift chamber devices [5].

Figure 2 explains the basic geometry of the MSE 2D-arrays we have studied. A part of a layout is shown, which was realized by using a standard CAD program. We chose the original proportional counter design as the starting point for our study, in which, for the first time, such structures are used as a discharge device. Hence, the geometry was basically maintained but some modifications were performed. For example, for the usage as a detector a readout of a single anode or a group of a few anodes is essential. This aspect becomes negligible, when the arrays are used as a discharge device.

The MSE 2D-arrays consist of parallel metal strips with a length of about 3 cm, which are bonded to a substrate and form, alternatively, anodes and cathodes. Between 40 and 270 strips with elliptically formed ends and with a width ranging from 50 to 400 mm act as cathodes. An area with a width of about 4 mm and a length of 40 mm, where the electrical contact with the mounting is established, links together all cathodes, so that a comb-like structure originates. A similar structure is formed by the anodes, but they are only between 12 and 25 mm wide. The anodes end with a circle 50 mm in diameter. The gap d between the anodes and cathodes was varied from 50 to 250 mm.

The initial step of an electrical breakdown process is an electron avalanche. An avalanche process starts with so-called seed electrons that are always present, e.g. due to cosmic rays. For an effective electron avalanche process a sufficiently strong electrical field is essential, because the seed electrons have to be accelerated in their mean free path gaining enough energy to ionize gas molecules via impact ionization [6]. Therefore, the width of the anodes has to be very small, i.e. in the range of a few micrometers. We chose 12 mm as the lower limit as a consequence of cost, because smaller structures are a large challenge in the manufacturing process.

A problem well known from the operation of microstrip structures as position sensitive gas detectors is the occurrence of sparks mainly at the ends of the electrodes, specifically at the cathode ends due to the strong electrical field in this region. Therefore, a lot of work have been done by many groups on optimizing the shape of the ends [7], [8]. The goal was to achieve a higher electrical field at the parallel sections of the electrodes than at the ends. For this, sharp edges have to be avoided in any case, but two additional aspects are important. Generally, an electrical field becomes weaker, when firstly the distance of two conductors with an opposite potential is increased and when secondly the radius of curvatures is chosen not too small. Tests with cathodes, which have semi-circular ends, revealed that sparking still occurs. But it was also observed that after sparking had taken place the ends became elliptically due to metal vaporization [9]. As a result, it was reasonable to design the ends of the cathodes elliptically from the start to minimize sparking, which is realized in our design. For similar reasons the anodes end with a small circle, likewise based on empirical results. We also increased the distance between the electrode ends and the adjacent leads to 2d.

At low pressures up to about 4000 Pa (30 Torr) it is possible to generate large-area homogeneous discharges with a low power consumption of the 2D-arrays. Therefore, the voltage drop is relatively small when breakdown occurs and the electrodes can be regarded as fully decoupled. However, this behavior changes, when the pressure is increased. Then both the breakdown voltage and the current increase. Depending on the design of the current lead the voltage distribution over the whole structure will be influenced. As a consequence, breakdown often occurs merely in a limited area of the array. Therefore, it is necessary to develop an array design which allows a creation of a large-area discharge as a combination of a great number of microdischarges decoupled with reference to their power consumption. The principle of the electrically decoupled microdischarges is illustrated in the right part of figure 3. Here, after breakdown has taken place at one location the discharge current does not influence the voltage applied to the adjacent electrodes. In contrast, the left part of figure 3 shows the situation of electrically coupled electrodes. The decoupling of the electrodes is achieved by adding an individual resistor to each cathode (see right part of figure 3).

The 2D-arrays discussed in this paper were standard photomasks fabricated by Photronics MZD GmbH, Dresden. Photomasks are commonly used in the semiconductor industry to transfer the circuit patterns onto wafers during the fabrication of integrated circuits. Hence, the mask-making process is a standardized process. As a result, the number of applicable materials for the manufacture of MSE 2D-arrays was limited for now. In particular, the only metal available for the electrodes was chromium. Hence, using e-beam technology the arrays were made of borosilicate or sodalime glass substrates on which a 0.1 mm chromium layer was evaporated. On a 4´´x 4´´ substrate four arrays could be manufactured, yielding a total array size of about 5 x 5 cm². The thickness of the substrate was 1.5 mm and 2.3 mm, respectively. Because of the area needed for the electrical contact the active size of the arrays , i.e. the maximal size of the discharge was about 4 x 3 cm².
 

Figure 4 schematically shows the experimental setup for the studies of the MSE 2D-arrays as a novel electrical discharge device. All measurements described in this paper were performed in a cubical chamber made of stainless steel with a total volume of about 1.5 litres, in which the array was placed by means of a mounting made of PVC. The very thin chromium electrodes do not allow conventional soldering. Therefore, the electrical contact had to be established by a modified spring contact. The special design of the mounting ensured that no sparking could occur at the contact points. The reaction chamber was evacuated to a pressure p £ 0.01 Pa (10^-4 Torr) before each measurement. To ensure an oil-free environment the employed system consisted of a turbo-drag-pump and a dry diaphragm vacuum pump. When evacuating the reactor the gas pressure was measured using a pirani or a cold cathode gauge, respectively. When adding the desired gas or gas mixture a capacitance manometer was used for a pressure measurement independent of the working gas composition. For all experiments, except for the LIF measurements, which were performed under static conditions, a gas flow rate between 10 and 50 sccm was set up by means of mass flow controller, when the desired pressure had been attained. Three gases could be adjusted simultaneously using a multichannel gas flow controller unit. For a fundamental characterization of the discharge, i.e. the determination of the range of the working pressure in dependence of the electrode gap and width, the general optical appearance, and the current-voltage-characteristics, experiments were carried out in the following gases with a purity of more than 99.99 %: helium, neon, argon, krypton, xenon and nitrogen. Besides, experiments with water vapor were performed.

For the operation with direct current (DC) a high voltage supply (Stanford Research Systems model PS 350) was used, which could provide voltages up to 5 kV of positive or negative polarity either. A ballast resistor was employed in series to limit the discharge current. For determining the current-voltage-characteristics, i.e. the sustaining voltage of the discharge, a digital multimeter in parallel connection was used. The current value was provided by the power supply.

For quantitative studies the experimental setup was outlined for the detection of nitrogen oxide (NO) via laser induced fluorescence (LIF) technique. Three windows allowed for optical access to the discharge. Tunable laser light with a wavelength around 226 nm probed the discharge volume a few millimeter above the MSE 2D-array surface. The emitted fluorescence light was detected under 90° by a photomultiplier (Valvo model XP2254B) mounted outside the reactor. The multiplier signal was monitored by means of a digital oscilloscope and was submitted to a gated integrator and boxcar averager (Stanford Research Systems model SR250), which was triggered by a photo diode. Finally the signal was transferred to a personal computer with a A/D-converter. For an absolute determination of the NO concentrations the system was calibrated with a pure NO gas filling. The experimental setup for the LIF measurements is described in detail elsewhere [10].

4.1 Optical characteristics of the discharge

For all gases, a uniform light emission from a region just above the array is observed. The discharge covers the whole active array surface with a typical thickness of approximately 0.5–1.5 mm. The thickness decreases with increasing pressure. Depending on the gas used and its pressure the discharge emits light with different color, owing to the electronic excitation energies of the gas and the kinetic energy of the free electrons in the discharge. Optical emission spectroscopy (OES) measurements have been performed in helium, neon, argon, krypton, xenon, nitrogen and water vapor in order to obtain detailed information on the electron energies in the discharge [10]. A uniform discharge covering the whole active area of the array (4 x 3 cm²) could be generated up to 5320 Pa (40 Torr) in helium (pink) and neon (orange-red). In these two discharges exclusively atomic transitions of neutral species were monitored [10]. In argon (violet), krypton (dark violet), xenon (gray-white), nitrogen (violet) and water vapor (violet) large-area uniform discharges are stable up to 665-1330 Pa (5-10 Torr). Beside transitions in neutral atoms the emission spectra of the discharges in Ar, Kr and Xe also showed lines of Ar⁺, Xe⁺, and Kr⁺, indicating electron energies of about 16 eV [10]. When the pressure is increased beyond the maximum values mentioned above the area covered by the discharge decreases continuously with increasing pressure in all gases. The maximum pressure at which a discharge can be generated is about 13300 Pa (100 Torr) in helium and neon and about 5320 Pa (40 Torr) in the other gases.

4.2 Current-voltage characteristic of the discharge

The arrays were tested with both polarities applied to the cathodes or anodes, either. Regarding the upper limit in pressure and the uniformity of the discharge the best results were obtained with a negative voltage applied to cathodes (fig. 2). The voltages applied in order to generate the discharge ranged from -1500 to -2500 V. The sustaining voltage of the discharge U_S was in the range from -200 to -500 V depending on the array, the gas, and the gas pressure. The discharge current was between 0.1 and 2.0 mA. Hence, the discharge showed the electrical characteristics of a normal glow discharge. The current was limited to 2 mA, in order to avoid damage of the arrays at high currents due to heating. Considering this, the appropriate value of the ballast resistor (2 MW or 2.8 MW) was selected. The resistance of the discharge was estimated to lie between 0.1 and 6 MW from the values of the sustaining voltage and the discharge current .

Figure 5 shows typical current-voltage characteristics of a MSE 2D-array discharge in helium for various pressures ranging from 133 to 1330 Pa (1-10 Torr). The electrode gap d of the array was 154 mm and the applied voltage U = -2500 V. Thus, the strength of the electrical field E = U/d was 1.6 x 10⁵ V/cm. The cathodes were 400 mm wide, the width of the anodes was 12 mm. In the data displayed in figure 5 the sustaining voltage is between -200 and -310 V and the discharge current is in the range from 1 to 1.8 mA. The voltage necessary for sustaining the discharge becomes lower at higher pressures. The current-voltage characteristic has a positive slope, i.e. the discharge shows a resistive behavior in the studied pressure range. A simple estimate of the power consumption employing the relation P = U_S ×  I yields a value between 0.2 and 0.6 W for the power consumption of the whole array. Thus, the power density is in the range from 0.015 to 0.05 W/cm².

4.3 Decomposition of NO

In order to demonstrate the capability of the MSE 2D-arrays for pollution control, the removal of nitrogen oxide (NO) from N₂/NO mixtures was investigated. NO molecules are excited by laser light of a wavelength around 226 nm from the X ^2P ground state to the A²S⁺ excited electronic state. The NO concentration was determined from the intensity of the fluorescence light. This diagnosis technique works quantitatively only for gas pressures in the few Torr regime, because of the large absorption cross section of the fluorescence light resulting in self absorption. The measurements were performed under static conditions, and, therefore, the gas exchange in the reactor proceeds by diffusion only. The time dependent removal of NO by the MSE 2D-array discharge is shown in figure 6. An increasing LIF signal was observed, when the reaction chamber was filled with a mixture of 13.3 Pa (0.1 Torr) NO and 665 Pa (5 Torr) N₂. As the discharge started the LIF signal immediately began to decrease due to the decomposition of the NO molecules. The decomposition rate was about 1.1 x 10¹⁵ molecules/s. In this example the sustaining voltage was 570 V and the current amounted to 0.1 mA. Finally, as the reactor was evacuated the LIF signal returned back to a zero value.

From the rotational spectrum and the width of the absorption line the NO gas temperature, i.e. the discharge temperature, was determined. The temperature of the discharge remains close to ambient gas temperature [10].
 

The results reported in this paper are part of a comprehensive study on MSE 2D-arrays. They confirm the expectation that the arrays could act as a new efficient discharge device in a wide range of applications in plasma technology. In this paper the capability of MSE 2D-arrays was demonstrated in the field of pollution control. Detailed investigations have to follow now and are for the most part in progress. One important goal is to generate discharges with MSE 2D-arrays at higher pressures up to atmospheric pressure.

Until now, experiments with arrays originally designed for the use as gaseous proportional counters were performed, with chromium as electrode material and glass as substrate. In future work other metals, e.g. gold or copper, will be investigated as electrode material due to their low resistance. In addition, other substrates than glass will be tested, because the conductivity of the materials was found to be important in the original application as position sensitive detector. Here, ceramics are of special interest. Also, silicon has a high application potential since most of the micromachining processes are optimized for this material. While the use of polymers as substrates is interesting too, especially with regard to the surface treatment of non-planar substrates as mentioned in the introduction, it is still a large challenge for standard manufacturing techniques. Apart from optimizing the arrays with respect to the employed materials another important aspect is the use of other power supplies than DC, i.e. pulsed DC, AC, and RF, for the generation of electrical discharges with MSE 2D-arrays.
 

This work was supported by the Deutsche Bundesstiftung Umwelt (DBU), Germany.

References

Figure 2.
Basic geometry of the MSE 2D-arrays. The drawing is not to scale in the vertical axis. The arrays consist of parallel metal strips with a length of about 3 cm which form, alternatively, anodes and cathodes. Strips with elliptically formed ends and with a width ranging from 50 to 400 mm act as cathodes. All cathodes are linked together by an area for the electrical contact, so that a comb-like structure originates. A similar structure is formed by the anodes, but they are only between 12 and 25 mm wide. The anodes end with a circle 50 mm in diameter. The gap d was varied from 50 to 250 mm.

Figure 3.
Decoupling of the electrodes. Depending on the design of the current lead the voltage distribution over the whole structure will be influenced. Therefore, an array design is necessary where the electrodes are decoupled with reference to their power consumption. The decoupling is achieved by adding an individual resistor to each cathode (right part of the figure). Here, after breakdown has taken place at one location the discharge current does not influence the voltage applied to the adjacent electrodes. In contrast, the left part of the figure shows the situation of electrically coupled electrodes.

Figure 4.
Schematic of the experimental arrangement. Measurements were performed in a cubical chamber made of stainless steel. For monitoring the decomposition of NO tunable laser light with a wavelength around 226 nm probed the discharge volume a few millimeter above the MSE 2D-array surface. The fluorescence light emitted was detected under 90° by a photomultiplier (PMT). LIF measurements were performed under static conditions. For all other experiments a gas flow rate was set up by means of mass flow controller (MFC).

Figure 5.
Current-voltage characteristic of a MSE 2D-array with chromium electrodes in helium at different pressures between 133 and 1330 Pa (1-10 Torr). Electrode gap d = 154 mm, cathode width 400 µm, anode width 12 µm.

Figure 6.
Time-dependent decomposition of nitrogen oxide (NO) with a MSE 2D-array discharge. p(NO) = 13.3 Pa (0.1 Torr), p(N₂) = 665 Pa (5 Torr), U_S = 570 V, I = 0.1 mA. NO molecules are excited by laser light of a wavelength around 226 nm from the X ^2P ground state to the A²S⁺ excited electronic state. The NO concentration was determined from the intensity of the fluorescence light.

Auf diesem Webangebot gilt die Datenschutzerklärung der TU Braunschweig mit Ausnahme der Abschnitte VI, VII und VIII.