Goudket, Philippe Andre and Carter, Richard (2004) A study of multipacting in rectangular waveguide geometries. PhD thesis, Lancaster University.
Goudket_Thesis_Corrected_Final.pdf - Accepted Version
Available under License Creative Commons Attribution.
Download (11MB)
Abstract
Electron multipacting is a resonant process by which electrons build-up under the influence of a radio-frequency (RF) field. This process can occur in evacuated RF equipment such as the input coupler of accelerator cavities. The superconducting RF cavities designed by Cornell University, that are used in a number of synchrotron light sources including the DIAMOND Light Source, have had a history of vacuum breakdown in the CESR ring at Cornell with frequencies that would be inappropriate for a reliable synchrotron facility. This work aims to understand and correct the problem and ensure smooth operation of the cavities. The cause of the vacuum breakdown hindering the cavities’ operation at Cornell was identified as being multipactor in the rectangular input coupler waveguide. Prior studies carried out by R.L. Geng at Cornell University identified a number of solutions which he proposed to verify experimentally. Two series of experiments were carried out at Cornell University on short waveguide sections. The first session allowed us to observe, measure and attempt to suppress multipactor using techniques such as a longitudinal static magnetic bias field and a groove cut along the waveguide centreline. While the first technique was found to be quite effective, since a relatively weak 10G field was found to be sufficient to achieve complete multipactor suppression, the groove did not allow such total suppression of the multipactor though it did mitigate its effects. The second experimental session was designed to complement the first. The waveguide allowed the testing of other methods such as multiple grooves, a ridge in place of the groove, or surface coatings. The ridge proved to be as effective as a groove with regards to multipactor suppression, while multiple grooves proved to increase rather than reduce the total multipacting current. The waveguide could also be heated or cooled to study the effect of baking the surface as well as that of condensed gases. Surface coatings were tested and found to have the expected effect of lowering the multipactor current, but the surface areas covered as well as the vacuum quality achieved were insufficient to conclusively validate the use of coatings as a means of achieving multipactor-free operation of the coupler. Both series of experiments provided extensive measurements of electron currents at various locations on the waveguide and at a range of RF power levels; these were compared to simulations of multipactor developed using the MAGIC PIC code. The code results and the simulations were found to agree closely when using a secondary electron model including backscattered low energy electrons. The code was able to predict the effectiveness of a ridge, as well as agreeing with experimental observations. Instead of sharply defined multipactor bands as predicted by simple multipactor models, the multipactor current (above a certain power level) does not disappear completely even though it may show peaks and troughs for various values of the RF power. In conclusion, the magnetic bias is the only proven method to ensure multipactor-free operation of the CESR-type cavities. A ridge or a groove cut along the centreline of the waveguide could be a simple, passive way of limiting and retarding the effects of multipactor in the input coupler, while coatings should certainly be considered, though more research is needed to fully validate the concept.