Abstract: In this study we quantitatively investigate for the first time the plasma characteristics of an argon gliding arc with a 3D model. The model is validated by comparison with available experimental data from literature and a reasonable agreement is obtained for the calculated gas temperature and electron density. A complete arc cycle is modeled from initial ignition to arc decay. We investigate how the plasma characteristics, i.e., the electron temperature, gas temperature,
reduced electric field, and the densities of electrons, Ar+ and Ar2+ ions and Ar(4s) excited states, vary over one complete arc cycle, including their behavior in the discharge and post-discharge. These plasma characteristics exhibit a different evolution over one arc cycle, indicating that either the active discharge stage or the post-discharge stage can be beneficial for certain applications.

Abstract: In traditional capacitively coupled plasmas, the discharge can be described by an electrostatic model, in which the Poisson equation is employed to determine the electrostatic electric field. However, current plasma reactors are much larger and driven at a much higher frequency. If the excitation wavelength k in the plasma becomes comparable to the electrode radius, and the plasma skin depth d becomes comparable to the electrode spacing, the electromagnetic (EM) effects will become significant and compromise the plasma uniformity. In this regime, capacitive discharges have to be described by an EM model, i.e., the full set of Maxwells equations should be solved to address the EM effects. This paper gives an overview of the theory, simulation and experiments that have recently been carried out to understand these effects, which cause major uniformity problems in plasma processing for microelectronics and flat panel display industries. Furthermore, some methods for improving the plasma uniformity are also described and compared.

Abstract: A one-dimensional, implicit particle-in-cell Monte Carlo collision model is used to simulate the plasma kinetic properties at a steady state in a parallel-plate direct current argon glow microdischarge under various operating conditions, such as driving voltage (301000 V) and gap size (101000 µm) at atmospheric pressure. First, a comparison between rf and dc modes is shown for the same pressure, driving voltage and gap spacing. Furthermore, the effect of gap size scaling (in the range of 101000 µm) on the breakdown voltage, peak electron density and peak electron current density at the breakdown voltage is examined. The breakdown voltage is lower than 150 V in all gaps considered. The microdischarge is found to have a neutral bulk plasma region and a cathode sheath region with size varying with the applied voltage and the discharge gap. In our calculations, the electron and ion densities are of the order of 10181023 m−3, which is in the glow discharge limit, as the ionization degree is lower than 1% . The electron energy distribution function shows a two-energy group distribution at a gap of 10 µm and a three-energy group distribution at larger gaps such as 200 µm and 1000 µm, emphasizing the importance of the gap spacing in dc microdischarges.

Abstract: In this study, we investigated the diffusion of H-atoms to the subsurface and their further diffusion into the bulk of a Ni(111) crystal by means of density functional theory calculations in the context of thermal and plasma-assisted catalysis. The H-atoms at the surface can originate from the dissociative adsorption of H2 or CH4 molecules, determining the surface H-coverage. When a threshold H-coverage is passed, corresponding to 1.00 ML for the crystalline Ni(111) surface, the surface-bound H-atoms start to diffuse to the subsurface. A similar threshold coverage is observed for the interstitial H-coverage. Once the interstitial sites are filled up with a coverage above 1.00 ML of H, dissolution of interstitial H-atoms to the layer below the interstitial sites will be initiated. Hence, by applying a high pressure or inducing a reactive plasma and high temperature, increasing the H-flux to the surface, a large amount of hydrogen can diffuse in a crystalline metal like Ni and can be absorbed. The formation of metal hydride may modify the entire reaction kinetics of the system. Equivalently, the H-atoms in the bulk can easily go back to the surface and release a large amount of heat. In a plasma process, H-atoms are formed in the plasma, and therefore the energy barrier for dissociative adsorption is dismissed, thus allowing achievement of the threshold coverage without applying a high pressure as in a thermal process. As a result, depending on the crystal plane and type of metal, a large number of H-atoms can be dissolved (absorbed) in the metal catalyst, explaining the high efficiency of plasma-assisted catalytic reactions. Here, the mechanism of H-dissolution is established as a chemical identifier for the investigation of the reaction kinetics of a chemical process.

Abstract: A hybrid Monte Carlofluid model is applied to simulate the wafer-temperature-dependent etching of silicon with SF6 inductively coupled plasmas (ICP). The bulk plasma within the ICP reactor volume as well as the surface reactions occurring at the wafer are self-consistently described. The calculated etch rates are validated by experiments. The calculations and experiments are performed at two different wafer temperatures, i.e. 300 and 173 K, resembling conventional etching and cryoetching, respectively. In the case of cryoetching, a physisorbed SFx layer (x = 06) is formed on the wafer, which is negligible at room temperature, because of fast thermal desorption, However, even in the case of cryoetching, this layer can easily be disintegrated by low-energy ions, so it does not affect the etch rates. In the investigated pressure range of 19 Pa, the etch rate is always slightly higher at cryogenic conditions, both in the experiments and in the model, and this could be explained in the model due to a local cooling of the gas above the wafer, making the gas denser and increasing the flux of reactive neutrals, like F and F2, towards the wafer.

Abstract: In this work particle-in-cell/Monte Carlo collision simulations are performed to study the heating mechanism and plasma characteristics in direct current (dc) superposed radio-frequency (RF) capacitively coupled plasmas, operated both in single-frequency (SF) and dual-frequency (DF) regimes. An RF (60/2 MHz) source is applied on the bottom electrode to sustain the discharge, and a dc source is fixed on the top electrode. The heating mechanism appears to be very different in dc superposed SF and DF discharges. When only a single source of 60 MHz is applied, the plasma bulk region is reduced by the dc source, thus the ionization rate and hence the electron density decrease with rising dc voltage. However, when a DF source of 60 and 2 MHz is applied, the electron density can increase upon addition of a dc voltage, depending on the gap length and applied dc voltage. This is explained from the spatiotemporal ionization rates in the DF discharge. In fact, a completely different behavior is observed for the ionization rate in the two half-periods of the LF source. In the first LF half-period, the situation resembles the dc superposed SF discharge, and the reduced plasma bulk region due to the negative dc bias results in a very small effective discharge area and a low ionization rate. On the other hand, in the second half-period, the negative dc bias is to some extent counteracted by the LF voltage, and the sheath close to the dc electrode becomes particularly thin. Consequently, the amplitude of the high-frequency sheath oscillations at the top electrode is largely enhanced, while the LF sheath at the bottom electrode is in its expanding phase and can thus well confine the high-energy electrons. Therefore, the ionization rate increases considerably in this second LF half-period. Furthermore, in addition to the comparison between SF and DF discharges and the effect of gap length and dc voltage, the effect of secondary electrons is examined.

Abstract: A Monte Carlo model is utilized for studying the behavior of electrons in the afterglow of an analytical microsecond dc pulsed glow discharge. This model uses several quantities as input data, such as electric field and potential, ion flux at the cathode, the fast argon ion and atom impact ionization rates, slow electron density, the electrical characterization of the pulse (voltage and current profiles) and temperature profile. These quantities were obtained by earlier Monte Carlo fluid calculations for a pulsed discharge. Our goal is to study the behavior of the so-called Monte Carlo electrons (i.e., those electrons created at the cathode or by ionization collisions in the plasma which are followed by using the Monte Carlo model) from their origin to the moment when they are absorbed at the cell walls or when they have lost their energy by collisions (being transferred to the group of slow electrons) in the afterglow of the pulsed discharge. The thermalization of the electrons is a phenomenon where the electron-electron Coulomb collisions acquire a special importance. Indeed, in the afterglow the cross sections of the other electron reactions taken into account in the model are very low, because of the very low electron energy. We study the electron energy distributions at several times during and after the pulse and at several positions in the plasma cell, focusing on the thermalization and on the behavior of the electrons in the afterglow. Also, the time evolution of the rates of the various collision processes, the average electron energy, the densities of Monte Carlo and slow electrons and the ionization degree are investigated.

Abstract: In this paper, the effects of prolonged bias-enhanced nucleation (prolonged BEN) on the growth mechanisms of diamond are investigated by molecular dynamics (MD) and combined MD-Metropolis Monte Carlo (MD-MMC) simulations. First, cumulative impacts of CxHy+ and Hx+ on an a-C:H/nanodiamond composite were simulated; second, nonconsecutive impacts of the dominant ions were simulated in order to understand the observed phenomena in more detail. As stated in the existing literature, the growth of diamond structures during prolonged BEN is a process that takes place below the surface of the growing film. The investigation of the penetration behavior of CxHy+ and Hx+ species shows that the carbon-containing ions remain trapped within this amorphous phase where they dominate mechanisms like precipitation of sp3 carbon clusters. The H+ ions, however, penetrate into the crystalline phase at high bias voltages (>100 V), destroying the perfect diamond structure. The experimentally measured reduction of grain sizes at high bias voltage, reported in the literature, might thus be related to penetrating H+ ions. Furthermore, the CxHy+ ions are found to be the most efficient sputtering agents, preventing the build up of defective material.

Abstract: We have computationally investigated the introduction of copper elemental particles in an inductively coupled plasma torch connected to a sampling cone, including for the first time the ionization of the sample. The sample is inserted as liquid particles, which are followed inside the entire torch, i.e., from the injector inlet up to the ionization and reaching the sampler. The spatial position of the ion clouds inside the torch as well as detailed information on the copper species fluxes at the position of the sampler orifice and the exhausts of the torch are provided. The effect of on- and off-axis injection is studied. We clearly show that the ion clouds of on-axis injected material are located closer to the sampler with less radial diffusion. This guarantees a higher transport efficiency through the sampler cone. Moreover, our model reveals the optimum ranges of applied power and flow rates, which ensure the proper position of ion clouds inside the torch, i.e., close enough to the sampler to increase the fraction that can enter the mass spectrometer and with minimum loss of material toward the exhausts as well as a sufficiently high plasma temperature for efficient ionization.

Abstract: The modelling of a gliding arc discharge (GAD) is studied by means of the quasineutral (QN) plasma modelling approach. The model is first evaluated for reliability and proper description of a gliding arc discharge at atmospheric pressure, by comparing with a more elaborate non-quasineutral (NQN) plasma model in two different geometries – a 2D axisymmetric and a Cartesian geometry. The NQN model is considered as a reference, since it provides a continuous self-consistent plasma description, including the near electrode regions. In general, the results of the QN model agree very well with those obtained from the NQN model. The small differences between both models are attributed to the approximations in the derivation of the QN model. The use of the QN model provides a substantial reduction of the computation time compared to the NQN model, which is crucial for the development of more complex models in three dimensions or with complicated chemistries. The latter is illustrated for (i) a reverse vortex flow(RVF) GAD in argon, and (ii) a GAD in CO2. The RVF discharge is modelled in three dimensions and the effect of the turbulent heat transport on the plasma and gas characteristics is
discussed. The GAD model in CO2 is in a 1D geometry with axial symmetry and provides results for the time evolution of the electron, gas and vibrational temperature of CO2, as well as for the molar fractions of the different species.

Abstract: A two-dimensional self-consistent fluid model combined with the full set of Maxwell equations is employed to investigate an Ar/CF4 capacitively coupled plasma, focusing on the phase-shift effect on the plasma characteristics at various frequencies and gas mixture ratios. When the discharge is sustained by a single frequency at 13.56 MHz in an Ar/CF4 mixture with a ratio of 0.9/0.1, no obvious difference is detected between the electron densities obtained in the so-called electrostatic model (with only the static electric fields taken into account) and the electromagnetic model (which includes the electromagnetic effects). However, as the frequency increases to 60 and 100 MHz, the difference becomes distinct, due to the significant influence of the electromagnetic effects. The phase-shift effect on the plasma radial uniformity has also been investigated in a dual frequency discharge, i.e. when the top driven source is switched on with a phase difference phiv ranging from 0 to π, in the frequency range 13.56100 MHz. At low concentration of CF4 (10%), Ar+ ions are the major positive ions in the entire range of frequencies. When the frequency is low, i.e. 13.56 MHz, the Ar+ density exhibits an off-axis peak at phiv = 0 due to the edge effect, and a better uniformity caused by the phase-shift modulation is obtained at phiv = π. At 60 MHz, the Ar+ density varies from edge-peaked at phiv = 0 to uniform (i.e. at phiv = 0.53π), and finally at phiv = π, a broad maximum is observed at the centre due to the standing-wave effect. As the frequency increases to 100 MHz, the best radial uniformity is reached at 0.25π, and the maximum moves again towards the radial wall in the reverse-phase case (phiv = π) due to the dominant skin effect. When the frequency is fixed at 100 MHz, the phase-shift control shows a different behaviour at a high concentration of CF4. For instance, the ${\rm CF}_3

Abstract: Particle-in-cell/Monte Carlo collision simulations, coupled with an external circuit, are used to investigate the behavior of pulsed dual-frequency (DF) capacitively coupled plasmas (CCPs). It is found that the phase shift θ between the high (or low) frequency source and the pulse modulation has a great influence on the ion density and the ionization rate. By pulsing the high frequency source, the time-averaged ion density shows a maximum when θ = 90∘. The time-averaged ion energy distribution functions (IEDFs) at the driven electrode, however, keep almost unchanged, illustrating the potential of pulsed DF-CCP for independent control of ion density (and flux) and ion energy. A detailed investigation of the temporal evolution of the plasma characteristics indicates that several high frequency harmonics can be excited at the initial stage of a pulse period by tuning the phase shift θ, and this gives rise to strong sheath oscillations, and therefore high ionization rates. For comparison, the pulsing of the low frequency source is also studied. In this case, the ion density changes slightly as a function of time, and the time-averaged ion density shows the same trend as in the HF modulation for different phase shifts θ. Moreover, the time-averaged IEDFs at the driven electrode can be modulated, showing the potential to reduce the maximum ion bombardment energy.