In the first part of the paper in a short review the history of the development of gas-filled high power switches is described, including all the special switching devices, which had been proposed to overcome specific shortcomings of the basic configurations of gas-filled devices like spark gaps, cold and hot cathode thyratrons and triggered vacuum tubes. The second part treats in detail the most serious problem for gas-filled switches to be still competitive with solid-state switches: Lifetime limitation by electrode erosion. The pseudospark discharge as low-pressure cold cathode device is chosen to illustrate the most important mechanisms, which contribute to electrode erosion or are related to cathode emission processes in general. As next subject of the manuscript some promising new developments in the area of gas-filled switches are described. Those obviously show that there is still a variety of applications for them, which guarantee a bright future. Based upon the results of the previous part in the last part, the pros and cons are summarized with the goal to comment the on-going process of substitution or not-substitution of gas-filled switches by solid-state devices. Finally the physical and technological limits are listed together with an outlook to future chances with the development of both types of high power switches.

At the GSI Helmholtzzentrum fuer Schwerionenforschung GmbH a new accelerator complex, called Facility for Antiproton and Ion Research (FAIR), is under construction. Its main components are the SIS100 and SIS300 heavy ion synchrotrons. To operate their injection/extraction kicker magnet systems, modulators with pulse-forming networks (PFNs) are necessary. The PFNs will be charged to a high voltage up to 70 kV and discharged via a high-voltage switch. The switch has to handle currents up to 6 kA, pulse durations up to 7 microseconds with an overall lifetime exceeding 10 8 shots. The repetition rate is about 4 Hz and a current rise rate of at least 4*10 10 A/s is required. The only commercially available switch in this parameter range is actually a multi-gap thyratron. As an alternative, a threegap pseudospark switch is under development at GSI. It combines the major advantages of the thyratron with its low stand-by power as a cold-cathode device, as well as its insensitivity to large current reversal. Like for the thyratron, the maximum hold-off voltage of a single gap pseudospark switch is limited to about 35 kV. For a reliable hold-off voltage of 70 kV, a three-gap system was designed. Test results with a first prototype switch of this design are reported. The prototype has demonstrated a voltage hold-off capability of more than 80 kV. The circuit of capacitive and resistive voltage dividers was optimized to improve the switch control, the delay and the jitter values. As trigger unit, a conventional high-dielectric trigger is used. With such a trigger unit a crucial issue for minimum delay breakdown still remains the plasma coupling between the different gaps by drift spaces. Those drift spaces have to be designed carefully in order to minimize the internal delay of breakdown. An additional major issue is that the switch suffers from losses, which principally limit the lifetime of low-pressure gas discharge switches. A common way to minimize losses by anode dissipation is to integrate a so-called anode inductor. To see whether this technique can be used with a cold-cathode switch at low repetition rates and relatively large pulse lengths, preliminary tests with an anode inductor were performed.