Report on first plasma processing trial for a FRIB quarter-wave resonator cryomodule

Plasma processing has been shown to help mitigate degradation of the performance of superconducting radio-frequency cavities, providing an alternative to removal of cryomodules from the accelerator for refurbishment. Studies of plasma processing for quarter-wave resonators (QWRs) and half-wave resonators (HWRs) are underway at the Facility for Rare Isotope Beams (FRIB), where a total of 324 such resonators are presently in operation. Plasma processing tests were done on several QWRs using the fundamental power coupler (FPC) to drive the plasma, with promising results. Driving the plasma with a higher-order mode allows for less mismatch at the FPC and higher plasma density. The first plasma processing trial for FRIB QWRs in a cryomodule was conducted in January 2024. Cold tests of the cryomodule showed a significant reduction in field emission X-rays after plasma processing.

💡 Research Summary

The paper reports the first successful plasma‑processing trial performed on a full FRIB quarter‑wave resonator (QWR) cryomodule, demonstrating a viable pathway to mitigate field‑emission‑induced performance degradation without disassembling the cryomodule. FRIB operates 104 QWRs (80.5 MHz) and 220 half‑wave resonators; contamination and surface‑oxide buildup lead to increased field emission, limiting the achievable accelerating gradient. Earlier plasma‑processing experiments at FRIB used the fundamental power coupler (FPC) to ignite a neon‑oxygen plasma directly in individual cavities. However, at room temperature the FPC exhibits a severe impedance mismatch, and plasma often ignites in the coupler rather than the cavity, risking damage and limiting plasma density.

To overcome these challenges, the authors adopted higher‑order modes (HOMs) as the drive mechanism. Two HOMs were selected: a TEM 5λ/4 mode near 404 MHz (previously used in bench‑scale tests) and a second dipole mode near 605 MHz. Electromagnetic simulations (CST Microwave Studio) show that the 404 MHz mode concentrates electric fields in the three high‑field lobes at the bottom of the cavity, providing good surface coverage, while the 605 MHz mode offers higher achievable plasma density at the cost of a less uniform distribution. By keeping the FPC in its nominal installed position (rather than re‑adjusting it for stronger coupling), the experiment reproduced realistic operating conditions, albeit with increased mismatch that reduces the maximum attainable plasma density to roughly one‑third of that achievable with a custom‑adjusted coupler.

A dedicated mobile plasma‑processing system was built. It includes mass‑flow‑controlled argon (Ar) and Ar/O₂ (90 %/10 %) gas lines, a turbomolecular pump network with gate valves for safety, a residual‑gas analyzer for by‑product monitoring, and a 100 W solid‑state RF amplifier feeding the selected HOM through the FPC. Real‑time diagnostics comprise forward and reflected power sensors, a picoammeter on the FPC bias T, and a network analyzer tracking resonant frequency shifts. Three interlocks—high FPC current, low reverse power, and low transmitted power—automatically shut off the RF drive if coupler ignition occurs, protecting both the coupler and the cavity.

The processing sequence for each cavity was: (1) ramp RF power to ignite the plasma, (2) increase the drive frequency to the maximum shift (≈0.9 MHz for 404 MHz mode, ≈3.5 MHz for 605 MHz mode) to raise plasma density, (3) maintain continuous‑wave (CW) power for one hour, and (4) repeat five times per cavity (total 5–10 h). The plasma was generated at ≈58 mTorr (≈7.7 × 10⁻⁵ bar), a pressure chosen to balance ignition threshold and plasma density.

Cold‑test results after processing four of the eight cavities (those exhibiting measurable field‑emission X‑rays before treatment) showed a dramatic reduction in X‑ray emission at an accelerating gradient of 10 MV/m—levels fell to the background of the X‑ray detector, whereas pre‑treatment values ranged from 0.1 to 1 mR/h. No new multipacting barriers appeared, and the measured forward, reflected, and transmitted powers remained within nominal limits, indicating that the plasma did not introduce additional losses. Power‑dissipation analysis estimated total RF power consumption of ~5 W during processing, with roughly 1 W actually deposited into the plasma, confirming the efficiency of the HOM‑driven approach.

The authors conclude that HOM‑driven plasma processing can be scaled from bench‑scale cavity tests to full cryomodules, offering an in‑situ remediation technique that avoids costly cryomodule removal and re‑assembly. They note that systematic optimization of gas composition, flow rates, and processing duration remains to be performed, and that long‑term effects on intrinsic quality factor (Q₀) and maximum gradient (E_acc) should be investigated. Nonetheless, this work establishes a practical pathway for maintaining and extending the performance of large SRF linacs such as FRIB.