Comparative Studies of High-Gradient Rf and Dc Breakdowns

The CLIC project is based on normal-conducting high-gradient accelerating structures with an average accelerating gradient of 100 MV/m. The maximum achievable gradient in these structures is limited by the breakdown phenomenon. The physics of breakdowns is not yet fully understood quantitatively. A full knowledge could have strong impact on the design, material choice and construction of rf structures. Therefore, understanding breakdowns has great importance to reaching a gradient of 100MV/m with an acceptable breakdown probability. This thesis addresses the physics underlying the breakdown effect, focusing on a comparison of breakdowns in rf structures and in a dc spark setup. The dc system is simpler, easier to benchmark against simulations, with a faster turnaround time, but the relationship to rf breakdown must be established. To do so, an experimental approach based on optical diagnostics and electrical measurements methods was made. Following an introduction into the CLIC project, a general theoretical introduction into accelerating structures and the presentation of the state of the art breakdown model is given. Afterwards, the results of the breakdown experiments are presented together with the instruments and facilities used. As an initial measurement, spectra of breakdowns in copper rf structures and on copper dc samples were taken in the wavelength range from 300nm to 1000nm. The typical emission time measured is aro und 3 μs. These spectra both show a copper plasma emission spectrum with identified ionization levels up to CuIII and a continuum background which carries about 75% of the total integrated light emission. No significant contribution from other elements was found in rf nor dc. Plasma temperature calculations from these spectra show inconsistent results: depending on the choice of lines, the values spread between 0.3 eV and 1.3 eV. Therefore, the plasma is likely to be a non-LTE plasma. No qualitative differences between rf and dc were found with these experiments. Emerging from the spectroscopy studies, light emission from the spark gap in the dc setup was observed in the case when no breakdown was triggered. This light was identified as an optical transition radiation (OTR), emitted by the field-emitted electrons hitting the copper surface. The measured OTR light intensity was found to be usable to measure the field enhancement factor of the field emitter on the sample surface at high electric field gradients near the breakdown limit where electric measurements could not be applied because of the risk of destruction of the electronics. The same OTR light emission was found also for a 30GHz rf structure. In order to get more insight into the time development of the breakdown plasma, time-resolved spectra were taken in rf and dc. It was found that the light emission of both continuum and lines consistently separate into two part s: an about 10 ns wide first, sharp peak which has to be attributed to the breakdown trigger mechanism, followed by a long intensity build-up and decay with a peak after about 1 μs after the triggering of the breakdown. Directly after the first peak, the intensity drops to practically zero. At the wavelengths of copper emission lines, this intensity of the second long peak is superimposed by peaks of different width, depending on the ion species. It is concluded, that the processes during a breakdown are highly dynamic in plasma temperature and density, not only in time, but also in spatial distribution of the plasma. Time resolved emission line waveforms show the same overall structure in both rf and dc, again indicating strong similarity in the phenomenon. In addition, the electrical characteristics of the dc spark were measured. It was found that the resistance of the spark goes to zero within less than 10 ns, showing that the plasma is a very good conductor. This limits the amount of energy dissipated by the spark. 30GHz breakdowns revealed a flat distribution of energy dissipated by the breakdown at constant input energy, reaching up to full rf energy loss. This different behavior can be explained by energy losses due to acceleration and loss of electrons and ions, which in the dc case still contribute to the current through the plasma. This effect is therefore a secondary effect and does not implicate a difference in the physics of rf and dc breakdowns. Summarizing, it can be concluded that rf and dc breakdowns are similar in their underlying physics in the scope of the experiments presented here.