The Design And Applications Of Fluid Simulation For Low Pressure Radio-frequency Capacitively Coupled Plasmas

Radio-frequency capacitively coupled plasmas (RF-CCP) have a wide range of applications in the semiconductor manufacturing industry including plasma etching, film deposition, etc. Currently, plasmas serve as an indispensable, key technology for the microelectronic fabrication industry. In practice, the efficiency and quality of the etching and deposition are largely depending on plasma state and working parameters of plasma sources. The plasma state is determined by external factors, such as parameters of power sources, chamber geometry, and properties of working gas. It exhibits rich nonlinear phenomena full of complicated physicochemical processes over a multiscale of space and time, which brings us great difficulites and challenges on both the reactor design and experimental diagnostics. Effective use of numerical simulations can not only help us design plasma sources for industrial application both efficiently and economically, but also assist us scientifically in deep understanding of the complex plasma behaviors occurring in plasma sources. Aiming at simulation of complicated physicochemical processes in a real RF-CCP chamber geometry, the purpose of this dissertation is to develop a fluid-Monte Carlo (MC) hybrid model to study how the CCP operating parameters affect the plasma characteristics and the energy of extracted ions.In this dissertation, we first briefly review the background, recent advances, and challenges of CCP, and also the problems we face in Chapter1. The contents of Chapter2to Chapter7are presented as follows.In Chapter2, there is a thorough introduction about the multi-modules of the simulation platform including fluid dynamic module, electrostatic module, MC collision module, and chemistry module, etc. The Poisson's equation is required to resolve simultaneously with the fluid module in algorithm due to the special properties of plasma, while the chemistry module conducts one-way coupling with the fluid module by the source terms. MC module starts after solving the plasma density, neutral particle density, and electron temperature by the fluid module through the one-way coupling as well to achieve the ion energy distribution and ion angular distribution (IAD). Input/output Module is designed for users to define their own reactor geometry, sources parameters, and functional settings, etc. Those modules could meet the user requirements to a great extent. Chapter3and Chapter4show the simulation of Argon discharge by our CCP solver as is described above. And the effect on plasma spatial uniformity is studied by modulating the pressure/RF sources parameters and reactor geometry parameters (chamber materials, geometric size, way of RF sources couplings, etc.) respectively. It is shown that within a certain range of parameters, the plasma spatial uniformity could be improved by increasing the pressure. the high frequency, and decreasing the high frequency power. The result for the study of reactor geometry parameters shows that insulators at the edge of electrodes could suppress the boundary effect, and a better uniformity is obtained by insulating the chamber wall simultaneously. When increasing the discharge gap, the plasma density increases initially and then stabilizes, along with a better uniformity at the radical direction. It is also found by altering the RF couplings that, the plasma density distribution is mainly caused by the location of high frequency source.The simulation of the reactive CF4/Ar gas is presented in Chapter5for the study of the reactive gas discharge properties including the components and spatial distributions of the charged particles. It is observed that the negative ions (F-in a major proportion) are mostly confined to the bulk plasma region where the densities are much higher than electrons. Then it will generate an electric field reversal near the bulk-sheath interface. The electron density has local maxima at that area. As the concentration of CF4is higher, the mixture gas shows more electronegative features, such as the growth of the electron density peaks. The distributions of positive ions (Ar+and CF3+) have similar shapes, which is due to the physical factors.The comparation between the numerical simulations and experimental diagnosis on DF-CCP Ar discharge is presented in Chapter6to verify the reliability of the results. A novel floating double probe for spatial plasma density and electron temperature measurements are proposed. The comparison result shows a good agreement between the measurements and the simulations, which indicate that the asymmetric effect of the axial plasma density distribution is caused by the self-bias voltage and also the diffusion effect out of the discharge region. In radical direction, the peak density appears at the electrodes edges. The electron temperature is approximately uniform at the bulk plasma region. While out of the electrodes, the electron temperature is cooled down.In Chapter7, the IED and IAD in CF4/Ar mixture gas discharge are investigated by numerical and experimental method respectively. It is observed that the IED shapes a multi-peak distribution with several sub-peaks presenting between the two main peaks. The number of the peaks equals to the ratio of HF/LF. The width between the two energy peaks is mainly affected by the low frequency sources. Compared with CF3+. Ar+possesses higher proportion of low-energy particles for the frequent charge exchange collisions. The numerical result shows good agreements with the experimental measurement on the shapes and positions of the energy peaks. However, the multi-peak distributions are not detected in the experiments because the ratio for the two frequencies cannot be exactly an integer (or half-integer).