Rework Recipe Development, Analysis and Results of Select 9-Cell Cavities for LCLS-II

The SLAC National Accelerator Laboratory is currently constructing a major upgrade to its accelerator, the Linac Coherent Light Source II (LCLS-II). Several Department of Energy laboratories, including the Thomas Jefferson National Accelerator Facility (JLab) and Fermi National Accelerator Laboratory (FNAL), are collaborating in this project. The cryomodules for this project each consist of eight 1.3-GHz cavities produced by two vendors, Research Instruments GmbH in Germany (RI*) and Ettore Zanon S.p.a. in Italy (EZ*), using niobium cell material from Tokyo Denkai Co., Ltd. (TD) and Ningxia Orient Tantalum Industry Co., Ltd. (OTIC/NX)). During the initial production run, cavity performance from one of the vendors (Vendor A) was far below expectation. All the cavities had low Q0, later attributed to minimal EP as well as high-flux-trapping NX material, early quench behavior below 18 MV/m, with many having Q0 roll-off at 12-16 MV/m. Production was stopped multiple times over the following 6 months, with test batches of cavities being made to ascertain the root cause of the problem. The final root cause of the problem was found to be inappropriate grinding of the RF surface prior to welding which left normal conducting inclusions in the surface. In addition, most cavities showed internal/external weld spatter which required post weld grinding and a very rough surface from operating the electropolishing machine in an etching rather than polishing regime. All issues have been corrected on new cavities and rework is underway on the originally effected cavities. TIMELINE FOR VENDOR A REWORK LCLS-II PRODUCTION • October 2016 – First 8 cavities arrive with baseline 140 μm EP/800°C heat treatment recipe with lower than expected Q0, quench field and strong Q-slopes on some cavities. • November 2016 – Next 8 cavities unsorted NX 9-cells arrive with 900°C/200 μm recipe flux trapping was still an issue, and Q0 varied widely between test setups [1-4]. These new cavities also had low quench field and strong Q-slopes on some cavities. Production was halted before all cavities available were tested. Analysis of the EP showed large variation in the total current, higher than expected temperature, and extreme surface roughness. Subsequent evaluation showed wrong cathode geometry and voltage, which caused etching rather than polishing during the chemical processing [5]. • December 2016 – JLab staff on site to verify production specified EP cathode and parameters. Four 9 cells without helium vessel sent thought production using the new EP parameters. • January 2017 – All 4 test EP cavities were RF tested and showed good low field Q0, but three still had a strong Q0 slope between 12 and 18 MV/m, production shutdown again. • February to March 2017 – Full review and onsite investigation into the root cause of the underperforming cavity results. • Shiny inclusions found on the sidewall of three cavities which received 275 μm EP, suggesting normal conducting inclusions. • Walkthrough of production work stations showed multiple steps where more inclusions were found from inadequate cleaning of tooling. • Global grinding with non-compliant tooling was found which could smear rather than remove metallic inclusions embedded in the niobium parts. • A cavity received 100 μm extra EP and still had Q-slope. • March to April 2017 – Four cavities went through the developed rework recipe with onsite direction from Jefferson laboratory staff. • July 2017 – All 4 cavities passed the operating specification and three of the four cavities pass the vertical test specification rework was initiated on the 63 cavities and material affected by global grinding. • present– As of April 12th 2018, 26 cavities have been tested with a pass rate of 76% EFFECT ON QUENCH FIELD WITH EXTRA EP When three of the four EP test cavities sent in January failed their RF test with the strong Q-slope before quench, the three failed cavities were tested again after an additional chemistry to understand what was causing the Q-slope. Two cavities received a 10 m BCP, and one received 100 m EP along with a re-dope at FNAL. One of the two BCP cavities went up in quench field while another didn’t, suggesting if BCP was used more than 10 m removal would be required in a rework recipe. The 100 9th International Particle Accelerator Conference IPAC2018, Vancouver, BC, Canada JACoW Publishing ISBN: 978-3-95450-184-7 doi:10.18429/JACoW-IPAC2018-THPAL140