Simulation of ion optics using particle-in-cell and treecode methods

A 2-dimensional and axisymmetric ion thruster optics simulation is presented in this dissertation. The simulation uses the Particle-In-Cell (PIC) method for particle motion and potential field computation, and the Direct Simulation Monte Carlo method is used to model collisions between particles. A treecode potential field solver is also developed for several domains, including ion optics.; The PIC model is first tested for accuracy of the potential field solver. It is found that the fineness of the mesh applied to the domain strongly affects the error in the field, and that the particle mesh is generally not fine enough to achieve reasonable accuracy. Correct simulation domain lengths are determined both upstream and downstream of the ion optics. The upstream domain length must be sufficient to allow correct formation of the ion extraction surface near the screen grid. The downstream domain length must allow collection of the correct amount of accelerator grid current.; The simulation is then applied to the NEXT ion engine to model the performance and life of the thruster. Apertures are modeled at various locations on the thruster surface to give a clear picture of the variation of performance quantities. The results from these simulations are integrated to give performance for the entire thruster, resulting in good comparison to experimental data. The life of the thruster is also estimated before failure due to electron backstreaming. Three separate methods are used to estimate thruster life, and all give approximately the same result.; A meshless treecode potential solver is developed for a 1-D electron sheath, a 2-D box domain, an axisymmetric cylinder domain, and an axisymmetric ion optics domain. The treecode has not been applied to plasmas in general prior to this work, and specifically the treecode has not previously been used in an axisymmetric domain. Comparison is made to direct summation of particle forces and the PIC method in terms of timing and accuracy. The treecode performs well in the 1-D and 2-D domains, but issues with boundaries and the treecode Taylor expansion cause the treecode to give relatively poor results in the axisymmetric ion optics domain.