Table of contents

Non-thermal plasma processing methods have been shown to be effective for treating dilute concentrations of pollutants in large-volume atmospheric-pressure air streams. This paper presents results from basic experimental and theoretical studies aimed at identifying the main reactions responsible for the decomposition of four representative compounds: carbon tetrachloride, methylene chloride, trichloroethylene and methanol. Each of these compounds is shown to be decomposed by a different plasma species: electrons, nitrogen atoms, oxygen radicals and positive ions, respectively. By understanding what plasma species is responsible for the decomposition of a pollutant molecule, it is possible to establish the electrical power requirements of the plasma reactor and help identify the initial reactions that lead to the subsequent process chemistry. These studies are essential for predicting the scaling of the process to commercial size units.

Pulsed laser irradiation of a solid target involves many physical processes starting with the coupling of a high flux of laser radiation to the surface, continuing with particle release and with laser - plume interaction, and ending with gas-dynamic expansion of the plume. The primary processes include electronic excitation followed by heating, and the secondary processes concern expansion of the plume as it moves away from the target surface. Theory and experiment on problems related to the surface phenomena, the laser - plume interaction, and the gas-dynamic expansion are discussed here.

Theoretical approaches to the accurate dynamical calculation of cross sections and rate coefficients of neutral atom - diatom reactions are illustrated. Computational steps needing to be accomplished in order to tackle the study of some gas phase reactions are examined. By making reference to our work calculated properties of these reactions are discussed in detail to stress when and where theory has been successful in reproducing experimental findings and to evidence some new insights stimulated by the theoretical investigation.

The one-dimensional transport of the neutral reactive species in the planar reactor is studied theoretically to draw conclusions on the reactor performance at maximum etching rate. The etching rate is calculated as a function of the reactant flux entering the reactor for various flux densities of desorbing product from the etched surface. An influence of the mutual diffusion of reactant and product on the etching rate is shown together with respective concentrations of reactant and reaction product in the reactor. The theory is compared with experimental results corresponding to etching of aluminium by chlorine. A method of in situ monitoring of the etching rate uniformity is also presented.

In this paper, the plasma ignition process above a metallic surface submitted to UV laser irradiation is studied. An easy model based on the hypothesis of thermal equilibrium between ejected vapour and heated surface, and of a local thermodynamic equilibrium state of the vapour, is used to characterize the metallic vapour at the end of the laser pulse. Then the efficiency of the different elementary mechanisms liable to sustain or to prevent the ionization process in this medium is discussed depending on the laser power density. In this work, the calculations are applied to the case of the interaction between an excimer XeCl laser beam ( nm, ns) and titanium target. It is shown that the thermal heated metallic vapour is partially ionized and contains excited and singly ionized species at high densities ( atoms ). The electron temperature in this medium is found to be around 1 eV. The study of the ionization rise in the vapour evidences the important role played by the single-photon ionization process and the electron/ion inverse bremsstrahlung effect. For laser power densities above 100 MW (laser fluence of 2 J cm), the ionization level is found to increase before the laser pulse end, and a thermal evaporation regime is reached. As the laser power density exceeds 500 MW (fluence of 10 J ), an avalanche breakdown is liable to occur in the vapour before the pulse end and the plasma governs the evaporation mode. The results presented here are in good agreement with experimental observations and with results from more complex models reported in the literature.

By using emission and classical absorption spectroscopy with a continuum light source we have investigated dielectric barrier discharges in /NO and mixtures. The concentrations of NO, , , , and were measured inside the discharge and in the exhaust. In the discharge space resolved absorption spectroscopy was performed. In discharges with high content of NO or electron impact induced dissociation of NO and turned out to be important. In discharges in pure nitrogen we found emission from NO and O. This indicates a surface reaction of atomic nitrogen with chemisorbed oxygen and desorption of atomic oxygen. No indication for the presence of physisorbed NO was detected. A novel method for obtaining the electron distribution function from absolutely measured light intensities was applied and the results compared to solutions of the Boltzmann equation.

Time and space resolved spectroscopic studies of the molecular band emission from are performed in the plasma produced by irradiating a graphite target with m radiation from a Q-switched Nd:YAG laser. High-resolution spectra are recorded from points located at distances up to 15 mm from the target in the presence of ambient helium gas pressure. Depending on the laser irradiance, time of observation and position of the sampled volume of the plasma the features of the emission spectrum are found to change drastically. The vibrational temperature and population distribution in the different vibrational levels of molecules have been evaluated as a function of distance for different time delays and laser irradiance. It is also found that the vibrational temperature of molecules decreases with increasing helium pressure.

A remote plasma enhanced chemical vapour deposition (RPECVD) reactor has been developed to deposit silicon oxide films. It consists of a microwave discharge (created by surface waves) along a quartz tube and a capacitively coupled radiofrequency (RF) discharge on a planar electrode (substrate holder) perpendicular to the tube and facing the gas flow. Plasma diagnostics have been performed in Ar, , Ar - and Ar - He - discharges, at two different positions of the reactor. The densities of electrons (a few ), argon metastables (a few ) and oxygen atoms ( - ) have been determined as a function of different plasma parameters, using microwave interferometry and optical emission spectroscopy (self-absorption and actinometry) respectively. In the case of , the gas flow has little effect on the local equilibrium along the discharge tube due to fast quenching on the walls and quenching in the plasma bulk by slow electrons and oxygen molecules. In contrast, the O atom density profile is governed by the gas flow velocity due to a slow recombination probability on the walls. However it clearly appears that the O atom recombination probability is much larger in the microwave discharge region (due to ion bombardment and plasma heating of the walls) than in the afterglow region. We have also shown that the fraction of O atoms with respect to molecules is enhanced by using helium and/or argon dilution due to the production of O atoms by the quenching reactions of or metastables with . Comparing the densities of electrons, argon metastables and oxygen atoms and their respective rate constants for the reaction with silane , we have deduced that the plasma chemical kinetics leading to silicon oxide deposition can be summarized in a simple scheme: electrons dissociate into O atoms and produce which subsequently react with to enhance the O atom density. In the flowing afterglow, is almost entirely decomposed by O atoms, the direct electron impact dissociation being negligible except when applying an RF discharge in the substrate region.

A detailed plasma - surface interaction study has been conducted using mass spectrometry and optical emission spectroscopy for the plasma characterization, and quasi in situ XPS analyses to control the surface chemistry modifications. The experimental results clearly evidence the different etching behaviour of In and P in the - plasma environment. A signature of the In and P etching mechanisms is available from the diagnostic of the plasma phase through the detection of phosphine as the major etching product of P and of excited In atoms related to the In etching reaction. The observation of the In emission line at 451.1 nm indicates a probable decomposition in the discharge of the organoindium compound which is regarded as the etch product of the group III element. XPS reveals the presence of P - H, C - In, In - In - C and In - In - P as surface species, and allows us to quantify precisely the P surface depletion. A time dependent etching mechanism is shown, first suggested by the etched thickness measurements and further confirmed by both etch product signal intensity and surface stoichiometry evolution with the plasma exposure time. Mass spectrometric signal and emission line intensities monitored as a function of time indicate that the In etching mechanism is responsible for this situation. It is concluded that the reactive ion etching of InP is under the control of the removal mechanism of In. The combination of surface analysis and plasma diagnostics is shown to be capable of providing an understanding of the plasma - surface interaction.

A 1.6 kJ plasma focus source was demonstrated as an electron source for microlithography. Resolution better than was obtained. The total energy in the beam and the energy of electrons was estimated from the lithographs to be > 20 mJ per shot with energy > 20 keV and > 1 J with energy keV.

The radical was detected in the gas phase during the etching of both Si and substrates under reactive ion etching conditions in steady state and in the afterglow of pulsed plasmas, by laser-induced fluorescence (LIF). Spatially and temporally resolved measurements show that there is a major source of these radicals at, or very close to, the etched substrates, and that desorption of as one of the primary products is the most likely explanation. As the absolute concentration was not determined, it is not currently possible to say whether is a major etch product, although the observed signals were large. With an Si substrate, is produced both under steady-state RIE conditions (i.e. in the presence of ion bombardment) and in the afterglow of a pulsed discharge (i.e. by pure chemical etching by F atoms). In contrast, with an substrate, is only produced in the steady-state plasma. The net surface reaction probability of was found to be close to unity on the reactor walls. Some possible gas-phase reactions of are also discussed. The fluorescence lifetime of the excited state of was measured for the first time, giving a value of ns, in good agreement with theoretical estimates.

A detailed kinetic model is used to investigate the mechanisms for ionization, dissociation and atomic re-association in a low-pressure positive column. The approach is based on the self-consistent solutions to the electron Boltzmann equation coupled to a system of rate balance equations for the levels, the electronically excited states of and the and ions. The maintenance electric field is self-consistently determined from the continuity equations for electrons and ions. The model provides a satisfactory explanation of measurements conducted in these conditions, in the range p = 0.6 - 2.5 Torr and I = 10 - 100 mA, for the reduced electric field and the concentrations of N atoms and and states. The rate coefficients and are derived here for the two reactions leading to associative ionization by collisions between electronic metastables and , respectively. The dissociation due to the vibration - vibration (V - V) and vibration - translation (V - T) energy exchanges is shown to represent only a minor contribution for the total rate of dissociation, in opposition to previous studies, due to the effects of fast V - T exchanges associated with - N collisions. Finally, it is shown that the reaction does not constitute an effective depopulating mechanism of N atoms as most of the N atoms so created are reconverted to the N by collisions on the wall and quenching.

A homogeneous kinetic model of a low-pressure - positive column is developed. The model is based on the self-consistent solutions to the electron Boltzmann equation coupled to the rate balance equations for the vibrationally excited molecules (X ) and (X ), the electronically excited states of , and NO(), and species. Further, this set is still solved together with the continuity equations for the electrons and the main positive ions (, , , , ) in order to determine the maintenance reduced electric field. This formulation allows us to determine the concentrations of the various neutral and ionic species, the electron density and the vibrational temperatures, and , as a function of the gas pressure, discharge current, gas temperature, tube radius and fractional composition. The calculated results are shown to be in satisfactory agreement with published experimental data. The complex interplay kinetics is analysed in detail and the effects of the poor knowledge of some collisional and surface data on the results are discussed.

In argon plasmas excited at electron cyclotron resonance above multipolar magnetic field structures, ion density increases linearly with microwave input power but saturates as it gets near the critical density. This behaviour is observed at the three microwave frequencies investigated, namely 960 MHz, 2.45 GHz, and 5.85 GHz, as well as for different magnetic field configurations. The saturation density is independent of the atomic or molecular nature of the gas, as shown with Ar and . Expectedly, the ion density saturation value varies as the square of the excitation frequency, while the microwave input power required to reach saturation is proportional to the critical density. For a given multipolar magnetic field confinement, the electron temperature is shown to decrease with increasing excitation frequency. This result stems from the confinement of the fast electrons, which generate the plasma. The evolution of the F-atom concentration in discharges, as measured by actinometry, is observed to saturate with microwave input power at values depending on gas pressures at both the 2.45 GHz and 5.85 GHz excitation frequencies. Ion density and F-atom concentration saturate at distinct microwave input powers.

By suitably adding right- and left-hand circularly polarized helicon waves in a cylinder, it is possible, in principle, to form linearly polarized modes, contrary to the notion that whistler waves must be circularly polarized. The plane-polarized component is accompanied by a left-hand circularly polarized component which vanishes on axis but becomes important at large radii. The field lines of these two components and the energy deposition profile are computed for an illustrative case.

The time dependence of plasma parameters in a pulse-time-modulated electron cyclotron resonance plasma of and Ar source gases is measured. It is found that plasma produces a large quantity of negative ions during afterglow and decay times of electron density, electron temperature and sheath potential of plasma are much smaller than those of Ar plasma. The negative ions in plasma stay for a long time in the afterglow and they quickly become extinct when the microwave power is turned on. These characteristics suggest that the pulse modulation of plasma produces a large amount of negative ions and changes the flow of charged particles through the sheath region to the substrate surface, which enables us to improve highly anisotropic and charge-free poly-Si etching.

The growth of solid residues within PECVD (plasma enhanced chemical vapour deposition) reactors has been extensively studied because of its implications for wafer particle contamination and is often referred to as dusty plasmas. On dielectric CVD (DCVD) production systems the coating of chamber walls and vacuum exhaust line with residues addresses also the issue of system maintenance. A common solution consists of periodically cleaning the deposition chamber by ionizing a PFC (perfluoro-compound) gas such as , or . This generates free fluorine radicals that dry etch the residues deposited on chamber walls. However, because of limited fluorine radical lifetime, this clean process is not efficient in the vacuum exhaust line where residues accumulate.

We propose an active solution to address the issue of solid waste treatment on a production DCVD system. We review the particular case of silicon nitride deposition, which is one of the worst known processes in terms of particle generation. These considerations are also valid for silicon oxide, silicon oxynitride, silicon carbide and amorphous silicon deposition processes. Here we report on our investigation on the particle formation, composition and morphology within a PECVD chamber and the deposition of these particles on chamber walls and vacuum exhaust line. We describe a method to design an efficient precipitator that traps the particles immediately downstream of the deposition chamber. The trapping uses gravitational and electrostatic means. This system does not necessitate any disposal procedure because of its capability to perform an in situ plasma assisted clean, reactivating the effluent PFC gas from the processing chamber. Here, the system is referred to as downstream plasma apparatus (DPA).

We present a model of an ECWR discharge, where electrons are heated by resonant absorption of electron cyclotron waves induced by an induction coil and propagating along a stationary magnetic field. We concentrate on a cylindrical discharge geometry; the external magnetic field is perpendicular to the cylinder axis. Using a finite element scheme, the time and space dependence of the induced electromagnetic fields are calculated under the assumption of a homogeneous plasma density. From these, the power deposition in the discharge, and the resonance condition for the external magnetic field can be determined.

A planar high-density plasma, 22 cm in diameter and 9 cm in length, is produced by a 2.45 GHz microwave radiation of 500 W through small slot antennas in argon at 20 - 350 Pa without a magnetic field. Several types of azimuthal and radial standing wave mode pattern are observed in the optical emission from the plasma depending on the discharge conditions. The microwave field in the plasma measured by a movable antenna decreases exponentially in the axial direction from the quartz wall adjacent to the slot antennas, thus suggesting the propagation of surface waves in the r, directions. The measured azimuthal microwave field distributions and the optical emission pattern clearly show a mode transition of the standing surface wave from a mode to a mode when the pressure is decreased from 140 to 44 Pa at the constant power of 400 W. Here denotes the transverse magnetic mode of azimuthal mode number m and radial mode number n. A wave dispersion analysis based on a one-interface uniform-density model predicts these modes in a range of electron densities corresponding to those measured by a Langmuir probe in the experiment.

Models of the electric field have been developed using numerical integration for inductively coupled discharges of cylindrical geometry. However, in some low-power-density, low-pressure applications, it is necessary to know only when the plasma can no longer be considered as having a negligible effect on the electric field in the discharge. A simple expression for the electric field at low discharge power densities is derived for an inductively coupled discharge maintained by rf currents through cylindrical coils and the conditions of its validity.

Macroscopic models for the equilibrium of a three-component electronegative gas discharge are developed. Assuming the electrons and the negative ions to be in Boltzmann equilibrium, a positive ion ambipolar diffusion equation is derived. Such a discharge can consist of an electronegative core and may have electropositive edge regions, but the electropositive regions become small for the highly electronegative plasma considered here. In the parameter range for which the negative ions are Boltzmann, the electron density in the core is nearly uniform, allowing the nonlinear diffusion equation to be solved in terms of elliptic integrals. If the loss of positive ions to the walls dominates the recombination loss, a simpler parabolic solution can be obtained. If recombination loss dominates the loss to the walls, the assumption that the negative ions are in Boltzmann equilibrium is not justified, requiring coupled differential equations for positive and negative ions. Three parameter ranges are distinguished corresponding to a range in which a parabolic approximation is appropriate, a range for which the recombination significantly modifies the ion profiles, but the electron profile is essentially flat, and a range where the electron density variation influences the solution. The more complete solution of the coupled ion equations with the electrons in Boltzmann equilibrium, but not at constant density, is numerically obtained and compared with the more approximate solutions. The theoretical considerations are illustrated using a plane parallel discharge with chlorine feedstock gas of p = 30, 300 and 2000 mTorr and , corresponding to the three parameter regimes. A heuristic model is constructed which gives reasonably accurate values of the plasma parameters in regimes for which the parabolic profile is not adequate.

Calculations concerning motion and evaporation of copper particles injected into an inductively coupled thermal plasma are presented. A simplified model of the plasma is used to determine temperature and velocity distributions from conservation laws and Maxwell equations. Trajectories and evaporation rates of injected particles are calculated. The influence of reduced particle residence time in the plasma on the evaporation behaviour is discussed. Optical emission spectroscopy is applied to measure temperature distributions in the argon and the argon/copper plasma. Radial and axial atomic copper density distributions are derived from the spectroscopic data. A comparison between the results of modelling and spectroscopic measurements is made and limitations of the techniques are shown.

For your convenience the contents are listed here in order of low to high pressure.

Europhysics Sectional Conference on Atomic and Molecular Physics of Ionized Gases, Poprad, High Tatras, Slovak Republic, 27--30 August 1996

Accurate calculations of cross sections and rate coefficients of some atom--diatom reactions relevant to plasma chemistry (270-279) A Laganà, S Crocchianti, G Ochoa de Aspuru, A Riganelli and E Garcíi a

Heating effects and gas-dynamic expansion of the plasma plume produced by irradiating a solid with laser pulses (260-269) A Giardini Guidoni, R Kelly, A Mele and A Miotello

Transport of chemically active species in plasma reactors for etching (280-297) V Martisovits and M Zahoran

Identification of mechanisms for decomposition of air pollutants by non-thermal plasma processing (251-259) B M Penetrante, M C Hsiao, J N Bardsley, B T Merritt, G E Vogtlin, A Kuthi, C P Burkhart and J R Bayless

PAPERS

Influence of the applied field frequency on the characteristics of Ar and SF₆ diffusion plasmas sustained at electron cyclotron resonance above multipolar magnetic field structures (386-393) T Lagarde, Y Arnal and J Pelletier

Calculation of electromagnetic fields and resonance conditions in a cylindrical ECWR discharge (415-426) R Krimke and H M Urbassek

Generation and extinction characteristics of negative ions in pulse-time-modulated electron cyclotron resonance chlorine plasma (398-404) T Mieno and S Samukawa

Plane-polarized helicon waves (394-397) F F Chen

Modelling plasma discharges at high electronegativity (437-449) A J Lichtenberg, I G Kouznetsov, Y T Lee, M A Lieberman, I D Kaganovich and L D Tsendin

Spatial analysis of C₂ band emission from laser produced plasma (317-322) S S Harilal, R C Issac, C V Bindhu, V P N Nampoori and C P G Vallabhan

Surface modification and etch product detection during reactive ion etching of InP in CH₄H₂ plasma (334-342) Y Feurprier, Ch Cardinaud and G Turban

Laser-induced fluorescence detection of SiF₂ as a primary product of Si and SiO₂ reactive ion etching with CF₄ gas (349-360) G Cunge, P Chabert and J-P Booth

Phenomenology of a dual-mode microwave/RF discharge used for the deposition of silicon oxide thin layers (323-333) R Etemadi, C Godet and J Perrin

The electric field in an inductively coupled, low-power-density discharge with cylindrical coils (435-436) P N Barnes

Mode identification of surface waves excited in a planar microwave discharge (427-434) M Nagatsu, G Xu, I Ghanashev, M Kanoh and H Sugai

Electron and heavy particle kinetics in a low-pressure nitrogen glow discharge (361-372) V Guerra and J Loureiro

Self-consistent electron and heavy-particle kinetics in a low-pressure N₂ - O₂ glow discharge (373-385) V Guerra and J Loureiro

Growth, trapping and abatement of dielectric particles in PECVD systems (405-414) S Raoux, D Cheung, M Fodor, W N Taylor and K Fairbairn

Evaporation of copper powders in an inductively coupled thermal rf plasma - numerical modelling and spectroscopic measurements (450-459) P Buchner, H Ferfers, H Schubert and J Uhlenbusch

Electron lithography using a compact plasma focus (343-348) P Lee, X Feng, G X Zhang, M H Liu and S Lee

A contribution to the understanding of the plasma ignition mechanism above a metal target under UV laser irradiation (298-306) A L Thomann, C Boulmer-Leborgne and B Dubreuil

Classical absorption and emission spectroscopy of barrier discharges in N₂/NO and O₂/NO_x mixtures (307-316) I P Vinogradov and K Wiesemann