Table of contents

A 3.0 GHz pulsed microwave source operated at atmospheric pressure with a pulse power of 1.4 MW, a maximum repetition rate of 40 Hz, and a pulse length of 3.5 µs is experimentally studied with respect to the ability to remove NO_x from synthetic exhaust gases. Experiments in gas mixtures containing N₂/O₂/NO with typically 500 ppm NO are carried out. The discharge is embedded in a high-Q microwave resonator, which provides a reliable plasma ignition. Vortex flow is applied to the exhaust gas to improve gas treatment. Concentration measurements by Fourier transform infrared spectroscopy confirm an NO_x reduction of more than 90% in the case of N₂/NO mixtures. The admixture of oxygen lowers the reductive potential of the reactor, but NO_x reduction can still be observed up to 9% O₂ concentration. Coherent anti-Stokes Raman scattering technique is applied to measure the vibrational and rotational temperature of N₂. Gas temperatures of about 400 K are found, whilst the vibrational temperature is 3000-3500 K in pure N₂. The vibrational temperature drops to 1500 K when O₂ and/or NO are present. The randomly distributed relative frequency of occurrence of selected breakdown field intensities is measured by a calibrated, short linear-antenna. The breakdown field strength in pure N₂ amounts to 2.2×10⁶ V m^-1, a value that is reproducible within 2%. In the case of O₂ and/or NO admixture, the frequency distribution of the breakdown field strength scatters more and extends over a range from 3 to 8×10⁶ V m^-1.

The absorption of a low-power rf signal from various shell antennas was measured in a short plasma column produced by a microwave electron cyclotron resonance source. Both azimuthally asymmetric (straight and helical) and symmetric (m = 0) antennas as well as antennas with rotating field were provided to compare their efficiencies. A multi-peaked dependence of the load resistance on density, which is intrinsic for any antenna, was shown to result from the excitation of axially standing helicon modes. The magnitudes of the absorption peaks were shown to substantially depend on both the antenna design and its position relative to the plasma column. The double half-turn and m = 0 antennas were measured to have a low excitation efficiency. The left- and right-hand helical antennas were found to be of equal impedance. The excitation efficiency of the phased antenna was considerably higher in the case of generation of the right-hand (m = + 1) polarized waves. Increasing input power stimulated a transition to a self-sustained helicon discharge which could exist in discrete modes with densities corresponding to the absorption peaks. Gas pressure and matching conditions were shown to substantially affect the antenna coupling to various modes and absorption efficiency. Measured results on the antenna load impedance were compared with computations performed with two alternative models. The model regarding the excitation of the Trivelpiece-Gould (TG) waves yields a reasonable agreement with experimental data. The absorption mechanism is argued to originate in the mode conversion of helicon waves into TG waves. Results obtained with the other model, which neglects the TG waves, were found to be strongly contrary to the measurements.

The effect of the ion- and atom-induced secondary electron emission yields for both `clean' and `dirty' cathode surfaces is investigated by means of a hybrid model, for typical conditions used in analytical direct current glow discharges (i.e. a pressure of 50-100 Pa, a voltage of 600-1200 V, and an electrical current of 1-10 mA). The hybrid model consists of a number of Monte Carlo models for fast electrons, fast argon ions and atoms in the cathode dark space, and sputtered copper atoms, a fluid model for slow electrons and argon ions, and a heat transfer model to calculate the gas temperature. For clean surfaces, secondary electron emission is almost exclusively attributed to argon ions, at the conditions under study. For dirty surfaces, on the other hand, fast argon ions and atoms contribute each about 50% to secondary electron emission, at the same discharge conditions. A so-called `apparent' secondary electron emission yield (i.e. per bombarding ion) is determined for the range of conditions under study. This value for clean surfaces was found equal to 0.07 for argon on a copper cathode, at all conditions investigated; for dirty surfaces, this value was always higher than 0.07 and it strongly depends on the discharge conditions. With these data, current-voltage-pressure characteristics have been calculated for both clean and dirty surfaces, and compared to experimental data. The absolute current values differ by a factor of 1-1.6 between clean and dirty surfaces. However, both calculated currents show more or less the same rise with voltage as the experimental data, in spite of the different behaviour of secondary electron emission yields for clean and dirty surfaces as a function of voltage.

Spatially resolved optical emission spectroscopy has been used to investigate the sputtering and ionization of titanium in a DC planar magnetron using argon as a sputtering gas. Maps of argon and titanium atomic and ionic emission have been obtained. Characteristic emission patterns have been identified and explained by the sputtering of atoms from the magnetron target and their subsequent ionization in the dense plasma at the magnetron target. The inverse Abel transformation was used to reconstruct plasma emission profiles, revealing that the non-uniform erosion profile at the magnetron target leads to a non-uniformity in plasma properties some distance into the discharge. An enhanced metal ionic content has been found in the plasma on the axis of symmetry of the discharge, and this has been explained by an enhanced lateral diffusion of metal ions arising from the change in momentum of sputtered metal during collisional ionization by metastable argon atoms.

Rayleigh scattering is used in order to deduce the gas temperature from neutral density measurements inside a microwave discharge in argon. Rayleigh scattering is a powerful non-intrusive method which provides direct heavy particle density measurements with a very high spatial resolution. The gas pressure ranges from 1 to 40 torr and the microwave power ranges from 35 to 900 W. We show that temperature gradient is very sharp at the edge of the plasma column: T_g drops to nearly room temperature within 3 cm. In contrast, T_g is nearly constant in the bulk of the plasma region. The gas temperature is determined over a large range of power and pressure conditions. We show that the gas temperature in the centre of the tube, ranging from 300 to 2500 K, is a linear function of the averaged power density per unit of length. Analytical calculations are in good agreement with experimental results.

The study of a pulsed helicon discharge with a planar Langmuir probe is presented. An unexpected and unreported effect has been observed, i.e. a large current peak on the probe at the end of the pulse. The post-discharge has then been carefully studied and the temporal evolution of the plasma parameters during the pulsed discharge was obtained by analysing the probe characteristics. In the post-discharge, a relaxation of the plasma parameters (plasma density (n_e), plasma potential (V_p), electron temperature (T_e)) was observed. The difference between the decay times measured for each parameter enables one to explain the current peak on the probe at the beginning of the post-discharge.

Laser Thomson scattering (LTS) is shown to provide a useful means for measurements of negative ion density in a glow discharge plasma. Because the saturation energy density is so low that complete photo-detachment occurs at a very early stage of the probing laser pulse for the LTS measurement, the main part of the laser pulse is Thomson scattered by the photo-detached electrons as well as those already generated in the discharge. The energies of photo-detached electrons are below the energy difference between the photon energy and the electron affinity of the negative ions, and Thomson scattering from these appears below the difference energy. The principle has been tested using the second harmonic of a YAG laser light (532 nm) against an inductively coupled plasma operated in a mixture of Ar(80%)/O₂(20%) at a pressure of 20 mTorr, and yielded the negative ion density of 1.5×10¹⁷ m^-3, which amounted to 37% of the electron density of 4.1×10¹⁷ m^-3 or 27% of the negative charges (electrons and negative ions).

Modulated beam mass spectroscopy is applied as a quantitative method for the measurement of the atomic nitrogen concentration in an expanding remote plasma. The plasma is created in a quartz tube by an electromagnetic surface wave (the upstream pressure ranges between 10 and 20 mbar). The recombined plasma with its N radicals is expanded into a reaction chamber where the pressure is about 10^-3 mbar. In order to take into account the losses of atomic nitrogen by sticking and recombination at the inner walls of the mass spectrometer, a chopper has been inserted between the orifice of the mass spectrometer and the ionization chamber. Hence, the measured signal is separated into a beam and a background component, with different dependencies on the expansion properties. The beam-to-background ratio is found to depend on discharge parameters such as total pressure or gas mixture composition. A correction factor for the wall losses which is independent of these parameters allows the quantitative measurement of the N flux.

Electron and ion energy distribution functions and other plasma parameters such as plasma potential (V_p), electron temperature (T_e) and electron and ion number densities (n_e and n_i) in low-pressure CF₄ plasmas have been measured. The experiments were conducted in a GEC cell using an inductively coupled plasma device powered by a 13.56 MHz radiofrequency (rf) power source. The measurements were made at 300 W of input rf power at 10, 30 and 50 mTorr gas pressures. Langmuir probe measurements suggest that n_e, n_i and V_p remain constant over 60% of the central electrode area, beyond which they decrease. Within the limits of experimental error (±0.25 eV), T_e remains nearly constant over the electrode area and peaks towards the electrode edge before falling rapidly. T_e and V_p increase with a decrease in pressure. n_e and n_i are not affected as significantly as T_e or V_p by variation in the gas pressure. The electron energy distribution function measurements indicate a highly non-Maxwellian plasma. CF₃⁺ is the most dominant ion product of the plasma, followed by CF₂⁺ and CF⁺. Significant amounts of etch products, SiF_x/COF_x(x = 0-3), of the quartz window were also detected. The concentrations of CF₂⁺ and CF⁺ are much larger than that is possible from direct electron impact ionization of the parent gas. The cross-section data suggest that the direct electron impact ionization of fragment neutrals and negative ion production by electron attachment may be responsible for increase of the minor ions.

Fourier transform infrared spectroscopy (FTIR) has been used to characterize inductively coupled CF₄ plasmas in a GEC reference cell in situ. In examining these FTIR spectra, several assumptions and approximations of FTIR analysis are addressed. This includes the density dependence of cross-sections, non-linear effects in the addition of overlapping bands and the effect of spatial variations in density and temperature. This analysis demonstrates that temperatures extracted from FTIR spectra may provide a poor estimate of the true neutral plasma temperature. The FTIR spectra are dominated by unreacted CF₄, accounting for 40-60% of the gas products. The amount of CF₄ consumption is found to have a marked dependence on power, and is nearly independent of pressure in the range of 10-50 mTorr. Small amounts of C₂F₆ are observed at low power. Also observed are etching products from the quartz window - SiF₄, COF₂ and CO - which occur in approximately equal ratios and together account for 17-19% of the gas at 300 W and 6-9% of the gas at 100 W. The concentrations of these species are nearly independent of pressure. CF_x radicals are below the detection limit of this apparatus (~10¹³ cm^-3).

During reactive glow discharge sputtering of copper in an argon/nitrogen plasma, we noticed an abrupt change of the target voltage and the deposition rate when the nitrogen concentration in the plasma exceeds a critical value. To explain this behaviour, the target surface after reactive glow discharge sputtering was examined by x-ray photoelectron spectroscopy (XPS). An experimental arrangement was constructed that allows direct transfer of the glow discharge cathode to the XPS analysis chamber without air exposure. These XPS measurements revealed that several different chemical states of nitrogen are present in the layer that forms on the target surface. The relative concentration of these different states changes when the critical nitrogen concentration in the plasma is exceeded.

Silicon nitride films were deposited using an atmospheric pressure plasma source. The discharge was produced by flowing nitrogen and helium through two perforated metal electrodes that were driven by 13.56 MHz radio frequency power. Deposition occurred by mixing the plasma effluent with silane and directing the flow onto a rotating silicon wafer heated to between 100°C and 500°C. Film growth rates ranged from 90±10 to 1300±130 Å min^-1. Varying the N₂/SiH₄ feed ratio from 55.0 to 5.5 caused the film stochiometry to shift from SiN_1.45 to SiN_1.2. Minimum impurity concentrations of 0.04% carbon, 3.6% oxygen and 13.6% hydrogen were achieved at 500°C, and an N₂/SiH₄ feed ratio of 22.0. The growth rate increased with increasing silane and nitrogen partial pressures, but was invariant with respect to substrate temperature and rotational speed. The deposition rate also decreased sharply with distance from the plasma. These results combined with emission spectra taken of the afterglow suggest that gas-phase reactions between nitrogen atoms and silane play an important role in this process.

The aim of this paper is to present and compare the results of two numerical models of electron transport in the channel region of stationary plasma thrusters, namely a Monte Carlo simulation and a diffusion model. This paper extends the model of Degond et al (1998 Transp. Theory Stat. Phys.27 203) to the case where elastic and inelastic collisions against atoms, as well as collisions with the walls are considered. After describing the extensions, we show numerical results of both models and compare them in two distinct situations. Though discrepancies between mean quantities such as the mean energy can reach a factor 2 in some cases, we show that qualitative properties of the electron population can be obtained at a reduced CPU cost by using the diffusion model. We conclude by showing that the electron population cannot be assumed Maxwellian in general, but that this assumption would lead to reasonable values for the ionization frequency if the mean energy of the population is known with reasonable accuracy.

We utilize an array of related optical diagnostics to deliver quantitative description of each process involving the 15 lowest levels in neon, for reduced electric fields ranging from 3×10^-16 to 2×10^-16 V cm² (3-20 Td). Description of the kinetics between the 1s and 2p excited states (Paschen notation) has been obtained using CW laser collisionally induced fluorescence (i.e. fluorescence from a non-excited upper level). We have employed this technique in combination with the traditional optical emission and absorption spectroscopic studies to determine the rate coefficients describing electron-collisional excitation from the ground to the 1s and 2p states. These values, along with previously determined 1s-2p rate coefficients, have been used with current literature cross sections to provide information on the electron distribution function for energy bands of 2-8, 16.9-23.0 and 19.0-25.5 eV. Values of the electron temperatures obtained from our 1s coefficients were used for an extensive study of the ground state to 1s excitation functions. From 3 to 20 Td, the average values of the temperature corresponding to the 1s excitations ranged from 1.8 to 3.1 eV, while those from 2p excitation were in the range 1.6-2.9 eV. Our investigation of the bulk temperature using data for 1s-2p excitation suggests a scaling of ~0.5 for the published theoretical excitation function, which is supported by the new experimental cross sections of Boffard et al. A quantitative description of cascade to the 1s states is also presented, showing that at the top of our reduced field range, cascade from the 2p states contributes 29-35% of the total excitation rates from the ground state.