High Power Microwave Surface Flashover On The Vacuum Side Of RF Dielectric Window

Multipactor breakdown on the vacuum side of RF dielectric window is the major factor limiting transmission and radiation of high power microwave on the order of 1 GW. In order to meet the requirements of higher power capability in the RF dielectric window, this paper presents a systematic study of the mechanism of window breakdown which employs high power microwave.As the electron multipactor is a type of vacuum discharge caused by secondary electron emission, J. R. Vaughan's empirical formulas for secondary electron emission are adopted. By assuming the linear polarization plane wave incident perpendicularly onto the dielectric window, a statistical theory for describing the multipactor discharge is brought forward. In order to simulate accurately the phenomenon of multipactor electron discharge on the RF dielectric window, an improved secondary emission model is proposed. Moreover a novel Monte-Carlo (MC) method based on this model is established. After verifying the validity of this MC code, susceptibility diagrams under different materials are constructed, and an analytical estimate of these curves within the method of statistics is presented to explain their configurations.Employing the dynamic model of multipactor, the time-dependent physics of the multipactor electron discharge on the vacuum side of the RF dielectric window is investigated. The results show clearly that the saturation state of the multipactor is an oscillatory steady state, in which the number of secondary electrons, the normal electric field built on the dielectric surface, the mean incident energy of the primary electrons and the average secondary emission yield (SEY) oscillate at twice the RF frequency. By analyzing the time trace of the superimposed oscillation of the DC and RF electric fields, Lissajous behavior of the multipactor discharge on the RF dielectric window is discovered. The temporal relationship between the normal DC electric field and the parallel RF electric field is found to trace in Lissajous'figure with a frequency ratio 2:1. These Lissajous'figures have different configuration and moving direction which are determined by the normalized RF electric field to its frequency.The MC simulations of the RF dielectric window breakdown have also shown the detailed evolution of multipactor in phase space. Since the periodic oscillation of the RF electric field, the pinch effect of space charges in the direction normal to the dielectric surface and the swinging phenomenon in the direction parallel to the surface are observed, density modulation is also exist in this process. All these mechanisms constrain the space charge to the local where the electrons are emitted, so the growing discharge could not expand into a large area. In comparison with the differences in electron movements between the surface flashover caused by RF and unipolar voltages, RF surface flashover under certain condition is supposed to trace in different appearance of discharge track, this is in good agreemwnt with the experimental results.From the MC simulations, the power deposited onto the dielectric is examined to be on the order of 1% of the incident RF power. This deposited power is demonstrated to have a close relationship with the material characteristics. Both the particle simulation and the calculation carried out by analytical theory have proved the material (or coating) with smaller SEY, lower initial energy and higher first cross-over energy to have a lower level deposited power.Finally, a novel method for measuring the complex permittivity based on TM0mn modes is proposed in order to measure the dielectric property of the RF window. Rigorous analysis based on the Ritz-Galerkin and mode matching method is performed to obtain the field distribution in the closed resonant cavity, and the formulas for measuring the complex permittivity are deduced. As for the measurement of Q-factor, the inverse mapping technique is adopted, which make use of the complex S-parameters and fit circles to the Smith Chart. The experiment conducted on PE has shown a relative precision of this measurement. In practice, the measurement error for the relative permittivity is within 1% while it is 5%~10% for the loss tangent.