Several ion sources failed during the last months of TFTR operation. Four suffered arc chamber leaks which admitted SF6 into the source, and one of these had grid rail warpage which resulted in damage to a water cooled scraper.

The arc chamber leaks occurred at the boundary between the probe plate and either the bucket or interface plate. Vacuum seals at those locations consist of a mylar insulator sandwiched between two o-rings. Evidence of arcing was apparent at the edge of the insulator between every other magnet. At the failure point, the mylar had been punctured to the o-ring. Arc tracks were in evidence along the plasma facing surfaces of the probe plate leading to every other gap between magnets.

This evidence is suggestive of forces driving the arcs toward the interface between the various plates and toward the o-ring. The force remains active until the magnetic field changes direction just outside of the o-ring. Damage is correlated with high arc power operation and the failure to expeditiously extinguish the arc. However, if the o-rings had been located slightly more outboard, the seal would not have been breached. Operation with additional anode area has been suggested as a means to avoid arc creation.

One of the damaged sources had warped grid rails and damage was sustained to a nearby scraper. TFTR ion sources are a masked down version of the US Common Long Pulse Ion Source. Four rails on either end of the source are masked. Those under the mask were the most deformed (0.05" out of tolerance). Damage is believed to have occurred during 3 s pulses when 15 MJ were extracted under slightly overdense conditions. The rails were deflected back towards the arc chamber causing the edge beamlets to be directed away from the axis. Warping was induced by thermal gradients in the accelerator between the central heated sections and the cool rails under the mask which extract no beam.

*Work supported by USDoE contract DE-AC02-76-CHO3073.