Cleaned tray 1. I had been trying to optimize the clean steps in the Standby recipe (#2) by lowering the flow of NF3 and reducing etch time. Returned NF3 flows and etch times to values prior to seeing the arcing. Ran a test with no problems.