Contamination and conditioning of the prototype double spoke cryomodule for European Spallation Source
A. Miyazaki, H. Li, K. Fransson, K. Gajewski, L. Hermansson, R. Santiago Kern, R. Wedberg, R. Ruber
CContamination and conditioning of the prototype double spoke cryomodule forEuropean Spallation Source
A. Miyazaki, H. Li, K. Fransson, K. Gajewski, L. Hermansson, R. Santiago Kern, R. Wedberg, and R. Ruber
Uppsala University, Uppsala, Sweden
A superconducting Double Spoke Resonator (DSR) is the technology of choice in a low energysection of a high power proton linear accelerator. At the FREIA laboratory in Uppsala University,we have tested two DSRs in a prototype cryomodule for the European Spallation Source (ESS)project. It showed that the conditioning process of these cavity packages would be the key for theseries production tests. In this paper, we present the conditioning procedure that we developed, andalso describe the results with a special focus on the cross-contamination observed between two high-power couplers. This study defines a standard conditioning recipe for the series DSR production forESS and also for future similar projects in the world.
I. INTRODUCTION
Spoke cavities made of superconducting bulk nio-bium [1] are a promising accelerating structure for lin-ear proton and heavy ion accelerators [2–6] in the sec-tion of around half the speed of light c as reviewed inRef. [7]. The European Spallation Source (ESS) projecthas adopted 13 cryomodules, each containing 2 DoubleSpoke Resonators (DSRs), in its low energy section [8].This section receives protons at 90 MeV from a nor-mal conducting drift tube section, accelerates them to216 MeV and sends them to the superconducting ellipti-cal cavity sections. The ESS DSR is optimized to havethe maximum transit time factor at β = v/c = 0 . β = 0 . β = 0 . v is the speed of protons. This will be the firstDSR operated in a proton accelerator project, and thusits experimental study is of great importance for the fu-ture proton machines. Table I summarizes the main ESSDSR parameters relevant to this paper. parameter valuefrequency f (MHz) 352.21operating temperature (K) 2accelerating gradient E acc (MV/m) 9average power consumption P c (W) 2.5optimal β Q ext . × operation band-width (Hz) 2000TABLE I: Main parameters of the prototype ESS DSR In 2015, we performed low-power tests on two proto-type spoke cavities with helium jackets in HNOSS (Hor-izontal Nugget for Operation of Superconducting Sys-tems) at the FREIA (Facility for Research Instrumen-tation and Accelerator development) laboratory [9–11].After that, in 2017, one prototype DSR with a powercoupler was installed in HNOSS and was powered by a tetrode power station [12, 13]. In those tests, we devel-oped and validated the infrastructure and testing proto-cols. The cavities were tested individually in HNOSS;thus in these previous tests, we did not address phenom-ena occurring particularly in the cryomodule where twocavities might interfere with each other by vacuum con-tamination.The next milestone was to study two DSRs in a pro-totype cryomodule [14] with more optimized parametersbased on the lessons learned in 2017. After the highpower test in HNOSS, the DSR was sent to IPN Orsay,and there it was rinsed and assembled into a prototypecryomodule with another DSR. Two power couplers werepreconditioned in a dedicated conditioning bench andwere assembled as well. This prototype cryomodule wasshipped to the FREIA laboratory and was tested in 2019,in the bunker as shown in Fig. 1. The aim of the testincluded studying the conditioning procedure [15], accel-erating gradient, dynamic heat load, cold tuner, piezo ac-tuator, diagnostics, Low-Level Radio Frequency (LLRF),and safety interlocks [16, 17]. In this paper, we reporton the new results from the conditioning process of theDSRs, which were crucial when these DSRs shared thesame beam vacuum in the prototype cryomodule.
FIG. 1: Photograph of the prototype cryomodule (right) andthe valve box (left) in the FREIA laboratory. a r X i v : . [ phy s i c s . acc - ph ] M a y This report is organized as follows. Section II givesgeneral information of the conditioning procedures per-formed in the FREIA laboratory. In section III, we dis-cuss the conditioning of the high power couplers. Theconditioning process and its results are followed by ourfindings on cross-contamination of the two couplers. Wedescribe this newly observed cross-contamination withpromising mitigation proposals for the series productioncryomodules. Section IV is dedicated to the cavity con-ditioning. Remarkably, the cross-contamination was notobserved between cavities unlike in the case of the highpower couplers. In section V, we describe the influence ofthermal cycles on the required conditioning of the cou-plers and cavities. This is of practical importance forcryomodule commissioning after installation in the ac-celerator. Section VI represents conclusions.
II. GENERAL INFORMATION OFCONDITIONING PROCESSES
Sufficient conditioning of couplers and cavities are ofcrucial importance in order to properly and safely evalu-ate and operate cavities in a cryomodule. The dedicatedtests in the FREIA laboratory are particularly useful tolocalize potential issues before installing the cryomodulein the accelerator tunnel, in which individual tests of eachcryomodule will be practically difficult. Figure 2 showsthe schematic of the ESS cryomodule for DSR, in whichtwo cavities dressed with helium tanks are connectedwith a beam pipe, surrounded by a thermal shield andan isolation vacuum vessel. High power coaxial couplers,with single TiN-coated ceramic windows, are mountedfrom the bottom of the cavities. The ESS DSRs are notequipped with higher-order mode dampers. In this proto-type cryomodule testing, the beam vacuum was pumpedfrom one side through the beam port with one turbo-molecular pump. The major distinction from our lastreport on the high-power test [12] was that now the twocavity packages shared the same beam vacuum.The FREIA laboratory is equipped with two power sta-tions dedicated to the ESS project [18]. During the pro-totype cryomodule testing, one power station was avail-able while the other was under commissioning [19]. Theconditioning was performed by switching from one cavitypackage to the other with the available high power sta-tion, which outputs maximum 400 kW (duty ratio 4.5%)from two combined tetrode tubes. The RF power wastransported to the cryomodule via a coaxial and rectan-gular waveguide system with high-power circulators andloads equipped with water cooling at a precisely con-trolled temperature. Just below the cryomodule, thewaveguide mode was converted to the coaxial mode by adoorknob adapter. The stress tests of the power stationand the waveguide system were included in the prototypetesting program.In the last report on one cavity testing [12] in theHNOSS cryostat, we noticed a substantial dynamic heat power couplers vacuum pumping linecavitiesbeam pipe
FIG. 2: Schematic cross-sectional view of the DSR cryomod-ule [20] ( c (cid:13)
CNRS/IJCLab) load in the cavity compared to a vertical test result. Dur-ing the disassembly of this poorly performing cavity pack-age, it turned out that the ceramic window was sputteredby copper. It was thought that this could be due tocontamination in the power coupler produced during itspreconditioning in the dedicated bench and/or assemblyonto the cavity package, which gave rise to plasma forma-tion during the high power tests and degraded the cavityperformance. Since that test was the very first high-power experiment with the DSR cavity package, the con-ditioning procedure was conducted with preliminary pa-rameters. The issue could be mitigated by optimizing thecontrol parameters of coupler conditioning, and there-fore, in this prototype cryomodule testing we decided toprocess the coupler more carefully and thoroughly.The conditioning procedure that we developed in theFREIA laboratory was divided into coupler processingand cavity conditioning. First, the high power couplerswere conditioned at room temperature before the coolingwas started. Secondly, the couplers were reconditionedafter the cavities were cooled down, in which they weretested at both the 4.2 K and 2 K temperatures. Finally,electron activities in the cavities were conditioned at 2 Kto reach the required accelerating gradient of 9 MV/m.Every time a thermal cycle occurred, for example for ded-icated tests of the cryogenic system, these three stepswere strictly followed to recover the functionality of thecavities.Some instruments were used as diagnostics for the con-ditioning process. One X-ray detector was placed nextto the cryomodule to detect radiation generated by elec-tron activities. Each power coupler was equipped with avacuum gauge next to the ceramic window. The electronactivities around the power coupler were monitored bya biased electron pick-up antenna. Several arc detectorswere also installed but were only partly used for techni-cal reasons. All these instruments were integrated into acontrol unit for monitoring and fast interlock purposes.
III. HIGH POWER COUPLER CONDITIONINGA. Procedure of coupler conditioning
To evaluate the cavity performance after the condi-tioning, we need around 120 kW forward power sentto the cavity to reach the target accelerating gradientof 9 MV/m, if the cavity resonance frequency and thegenerator frequency are perfectly tuned. In the HNOSStest [12] we conditioned the cavity package up to 120 kWand then measured the performance by locking the fre-quency with a Self-Excited-Loop (SEL). However, in theexperiment reported in this article, it was an importantmilestone to assess the functionality of the closed-loopfeedback control of the digital LLRF system for the ac-celerator operation. This required a total power around350 kW to achieve a flat field level of 9 MV/m during theperiod corresponding to a beam bunch passage (2.86 ms) . Hence, the conditioning was carried out to the maxi-mum power of 400 kW and, therefore, the potential stressto the system, including power couplers and waveguides,was higher than the previous study.The couplers were processed by a standing wave condi-tion, in which avoiding to power the cavity at warm wascrucial. The power station was driven by an open loopat a generator frequency of 353 MHz, which for a cav-ity resonant frequency of 352 MHz, sufficiently exceedingthe cavity band-width, which is around 30 kHz at warmand around 2 kHz at cold. The cavity field was assuredto be zero by monitoring the power through a pick-upantenna.The diagnostic system is crucial to detect any suspi-cious activities around the coupler when it is processed byRF. Amongst others, the vacuum gauges installed next toeach power coupler’s ceramic window are the most sen-sitive and reliable instruments for software control, andmore importantly, hardware interlocks. For the couplerconditioning, X-ray activity was not observed due to thelow field level.Table II shows the baseline parameters of the couplerconditioning defined in this prototype cryomodule test-ing. The coupler started to be conditioned with a short-est pulse length of 20 µ s. At each pulse length, the powersent to the cavity package was increased from 60 dBm(1 kW) to 86 dBm (400 kW) by 0.1 dB/s. The powerwas held constant if the vacuum level exceeded the lowervacuum threshold v L to await the recovery of the vacuumlevel. When substantial outgassing was detected and thevacuum level reached the higher vacuum threshold v H ,the power was decreased by 3 dB until the recovery of Strictly speaking, the total power of 350 kW during the first300 µs is necessary to decrease the charging time of the cavity.A similar overhead is required for the rest of the pulse for beamloading, which was reflected back to the external load during ourtest without proton beam. the vacuum level to v L . Once the power reached themaximum of 86 dBm without substantial outgassing, thepulse length was extended and power was scanned againfrom the minimum of 60 dBm. Finally, the conditioningwas accomplished with several power cycles at the fullpulse length of 3.2 ms. The process of controlling pulselength and power levels was automatically conducted bya LabVIEW software code.The two vacuum threshold values v L and v H were man-ually adjusted around the typical values listed in Table IIso that the conditioning period was optimized. On topof these controlling vacuum values, a hardware interlockwas defined at v = 10 − mbar and cuts the RF immedi-ately in case of a substantial and sudden vacuum jump. parameter valueminimum power (dBm) 60maximum power (dBm) 86pulse length ( µ s) 20, 50, 100, 250, 5001000, 2000, 3200baseline vacuum at warm (mbar) 10 − baseline vacuum at cold (mbar) 10 − hardware interlock v (mbar) 10 − higher threshold v H (mbar) ∼ × − lower threshold v L (mbar) ∼ × − TABLE II: Baseline control parameters for the coupler con-ditioning
B. Coupler conditioning at warm
Figure 3 shows the evolution of one coupler condi-tioning at warm, where technical interventions are dis-regarded from this plot. As expected, the outgassinglevel rose when either the power level or the pulse lengthwere increased, and it slowly recovered in time. Afteraround 50 hours, the vacuum level was stabilized below5 × − mbar, which is comparable to the baseline vac-uum at warm. We judged that the coupler was fullyconditioned at that moment.We confirmed two power levels where the outgassing ismore substantial than at other levels:1. 74 dBm (25 kW) and2. 76 dBm (40 kW) particularly with a pulse lengthlonger than 50 µ s.These two power levels may correspond to local multi-pacting barriers on the coaxial coupler and the ceramicwindow. Apart from vacuum outgassing, we did not ob-serve a clear indication of electron activities with otherdiagnostics, such as the electron pick-up antenna.Although the outgassing was fully stabilized after thesingle coupler conditioning, an outgassing level of one or-der of magnitude higher than the baseline vacuum was f o r w a r d po w e r [ d B m ] v a c uu m [ m ba r ] -7 -6 elapsed time [h]0 5 10 15 20 25 30 35 40 45 s ] µ pu l s e l eng t h [ FIG. 3: Evolution of coupler conditioning process at warm,with applied forward RF power controlled by the software(top), degradation of vacuum level detected by the vacuumgauge next to the coupler (middle), and pulse length of theRF (bottom). observed when the coupler was powered again after theconditioning of the other coupler. This required yet an-other conditioning of the first coupler, which is discussedin the next section.
C. Cross-contamination at warm
During the prototype cryomodule testing, the cavitypackages were conditioned one by one in a common beamvacuum. Therefore, the multipactors removed from onecavity package could contaminate the other. Then, dur-ing the conditioning of the other cavity package, the firstcavity package could also be re-contaminated. The re-conditioning of the couplers was performed again oneafter another. It is worth mentioning that this cross-contamination was only observed during the warm cou-pler conditioning, and correspondingly, the recondition-ing was only required at warm. Figure 4 shows the reconditioning of the coupler atwarm recorded after conditioning of the other coupler.Comparison with Fig. 3 shows that the outgassing level,stabilized to below v L = 5 × − mbar, became ac-tive again. We confirmed this phenomenon by the vac-uum rise to v H , which was set at 10 − mbar in the soft-ware for a thorough conditioning, when the pulse lengthreached 1 ms at 2 h in Fig. 4. At that moment, we de-cided to recondition more carefully to avoid further cross-contamination, and reduced the pulsed length. This re-conditioning took around 25 h, corresponding to half theinitial conditioning time of 50 h. A third conditioningwas not necessary due to a rather stabilized outgassinglevel. This cross-contamination happened to both cou-plers and therefore the total conditioning time at warmwas (50 + 25) × ∼
150 hours. f o r w a r d po w e r [ d B m ] v a c uu m [ m ba r ] -7 -6 elapsed time [h]0 5 10 15 20 25 s ] µ pu l s e l eng t h [ FIG. 4: Evolution of coupler reconditioning process at warm,with applied forward RF power controlled by the software(top), degradation of vacuum level detected by the vacuumgauge next to the coupler (middle), and pulse length of theRF (bottom).
One hypothesis of the cause of rather severe cross-contamination is the pumping path of this prototype cry-omodule. No local pumping ports were placed near thecouplers, and only one turbo-molecular pump was con-nected through the beam port of one of the cavities witha rather long flexible tube between the cryomodule andthe pump. Depending on the conductance of the pump-ing port, part of the contamination can be captured bythe surface of the other coupler rather than removed fromthe cryomodule via the vacuum pump.
D. Coupler conditioning at cold
After the coupler warm conditioning, including the re-conditioning of the cross-contamination, had been ac-complished, the cryomodule was cooled down. The base-line vacuum level was around 10 − mbar thanks to thecryogenic pumping, which also reduced the outgassingdetected by vacuum gauges mounted next to the powercoupler. We tried cold conditioning at 4 K and 2 K forafter a few thermal cycles.Figure 5 shows the typical evolution of coupler condi-tioning at cold. The conditioning of one coupler at coldtook only around 2 hours, independently of the ambienttemperature of either 2 K or 4.2 K. The thorough con-ditioning at warm mitigated the outgassing during thiscold conditioning.Unlike the conditioning at warm, the cross-contamination issue was not observed during theconditioning at cold. This was probably because of localcryogenic pumping around the coupler. E. Conditioning period for one coupler
Apart from the cross-contamination reconditioning atwarm of 25 hours, the total time used for the coupler con-ditioning was 50 (warm) +2 (cold) = 52 h. This 52 h canbe used as a reference time of one coupler conditioningtime to compare with other experiments.In the single cavity study [12], we used 14 h for thecold conditioning while we used only 30 h for the warmconditioning; thus, it took 44 h in total. The warm con-ditioning was 12 h shorter than the test reported in thisarticle. In the previous test, the vacuum controlling al-gorithm was not identical to that used during the pro-totype cryomodule testing, and typical threshold param-eters were one order of magnitude higher than those inTable II. The 14 h for cold conditioning was just a de-fault minimum period of the algorithm and was unneces-sarily long. Obviously, the procedure was not optimizedfor the coupler conditioning inside the cryomodule. Thewarm conditioning must have been more thorough, andthe cold conditioning could have been shorter. The opti-mized conditioning parameters in the present report ledto a promising performance after the cavity conditioning.The required conditioning periods are the baseline ofone DSR in an ESS series cryomodule. In reality, thecross-contamination between two DSR packages takes anadditional 25 h for one coupler. It is of importance to f o r w a r d po w e r [ d B m ] v a c uu m [ m ba r ] -9 -8 elapsed time [h]0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 s ] µ pu l s e l eng t h [ FIG. 5: Evolution of coupler conditioning process at cold,with applied forward RF power controlled by the software(top), degradation of vacuum level detected by the vacuumgauge next to the coupler (middle), and pulse length of theRF (bottom). consider a possible mitigation of the total conditioningby parallel processing. This possibility is discussed inthe next section.
F. To improve the total processing time
The experimental results so far showed that the cou-pler conditioning particularly at warm would be a bot-tleneck in the project schedule together with the cross-contamination issue. After the prototype cryomoduletest, we developed two improvements in our system tobe deployed for the series cryomodule testing.1. The second high power station was commissionedfor simultaneous conditioning of the two cavities.2. One additional vacuum pump will be installed atthe other end of the beam port to increase thepumping capacity.Especially important is the second point, which will de-couple the vacuum in the two cavity packages due tothe low-conductance beam pipe between them. This willhelp in preventing cross-contamination when outgassingoccurs in one cavity package.The simultaneous conditioning could increase twice thepossible vacuum jump, which was already saturated atthe controlling vacuum threshold v H . Therefore, apartfrom the influence of the cross-contamination, simultane-ous powering with the additional pump does not guaran-tee a decrease in the total conditioning process by a factorof two. However, we can avoid switching the high poweramplifier system from one cavity to the other, and theautomatic software would not be interrupted. Therefore,we will gain on the gross conditioning time by reducingsuch intervals.Also, A DC biasing on the coupler antenna can po-tentially avoid the multipacting inside the high powercoupler. The series power coupler can be operated withthis DC bias option, but its deployment in the ESS accel-erator is not a baseline at the time of writing this report.The technical design for mitigating the coupler issues ob-served in our study must be carefully investigated withrespect to the balance between benefits and risks dueto additional complexities of mechanical and operationalaspects. IV. CAVITY CONDITIONINGA. Procedure of cavity conditioning
After the outgassing from the couplers was stabilizedat cold, the cavities were eventually powered to a cer-tain field level. The SEL was developed in the previoustests [9, 12], with which the resonant frequency of thecavity was automatically tracked. However, in this proto-type cryomodule testing we needed to test the generator-driven operation since the SEL function will not be de-ployed in the ESS accelerator. We powered the cavity byan open loop with a generator frequency adjusted to theresonant frequency of each cavity, enabled by the wideband-width of the cavity being around 2 kHz, as shownin Table I.The conditioning of a cavity was carried out with short(1.0 ms) and then long (3.2 ms) RF pulses. We startedprocessing with the short pulse for safety and slowly in-creased the accelerating gradient, which was estimatedby the power picked up through an antenna. Electronactivities were associated with the X-ray dose rate, andsometimes also with outgassing, and the field level waskept constant until stabilized.The maximum field at the long pulse of 3.2 ms to beconditioned for was decided by the criteria:1. up to the field at which a thermal quench happensor2. maximum 15 MV/m if the cavity does not quench.Unlike the coupler conditioning, the cavity processing inthe prototype cryomodule was manually performed and thoroughly checked due to a higher risk of over-poweringby an unexpected cavity behavior. In general, the be-havior of the cavities under high power RF were less pre-dictable than for the couplers, especially considering thatthe DSR powering is a novel experiment in the world.
B. Multipacting barriers
The multipacting barriers have been studied during thevertical tests [21] and the dedicated HNOSS tests [12].In the vertical tests, a characteristic drop of the qual-ity factor due to the multipacting barriers were observedbetween 4 and 8 MV/m. Consistently, the HNOSS testwith the power coupler showed three multipacting bands:4.5-4.8 MV/m, 5.2-5.7 MV/m, and 7.0-7.5 MV/m.Figure 6 shows a typical result from the cavity condi-tioning. Note that the small absolute scale of the radia-tion dose rate is not crucial for the discussion because theposition of the radiation monitor was 4 m from the cav-ity, off-axis from the beam axis by 1.5 m, and with theother cavity acting as a radiation shield between. Theconditioning of one cavity took around 6 hours for onecavity, of which around 2 hours were used for the shortpulses and the rest was with the full pulse length.The outgassing and radiation were observed at fieldlevels consistent with the multipacting bands, also seen inthe vertical tests and the HNOSS tests. Above 8 MV/m,one cavity (Fig. 6) showed a mild field emission that waseasily conditioned in short time. The other cavity suf-fered from a harder field emission around similar fieldlevels but was cleaned after a small thermal cycle up to50 K.Both cavities exceeded the target field level at9 MV/m. One cavity was thermally quenched at10.5 MV/m while the other reached 15 MV/m. The dy-namic heat load at 9 MV/m met the spec (2.5 W) in bothcavities [16, 17].Unlike the coupler conditioning at warm, but similarto the cold coupler conditioning, no cross-contaminationwas observed. This is probably because the removed gaswas captured by the same cavity surface through cryo-genic pumping and/or redistributed to a region geomet-rically more insensitive to multipacting.
C. Outgassing at 74 dBm and 76 dBm
We observed vacuum activities up to 3 × − mbar ata forward power of 74 dBm with the short pulses and at76 dBm with the long pulses (Fig. 6). The associated ra-diation levels were relatively low, indicating that no ener-getic electron activity was causing this outgassing. Thesetwo bands were identical to the power levels where we sawoutgassing during the coupler conditioning. They werecompletely conditioned at warm and no activity above3 × − mbar was observed during the coupler cold con-ditioning (Fig. 5) before the cavity conditioning process. f o r w a r d po w e r [ d B m ] [ M V / m ] a cc E v a c uu m [ m ba r ] -9 -8
10 elapsed time [h]0 1 2 3 4 5 S v / h ] µ r ad i a t i on [ FIG. 6: Evolution of cavity multipacting conditioning. Fromthe top, forward power, accelerating gradient, vacuum level,and radiation level are shown. The left side of the verticaldashed line at 2.3 h was conditioned with pulse length of1 ms while the right side was processed with 3.2 ms.
A difference from the coupler conditioning was theinput impedance of the cavity. The coupler was con-ditioned by a standing-wave produced by a frequencymismatch of around 1 MHz above the cavity resonantfrequency. The cavity was, however, processed by astill standing-wave in the coupler but from an over cou-pling condition with frequency matched within the cav-ity band-width. These different impedance conditionsresulted in slightly different standing-wave distributions.These two different standing-wave patterns gave rise toan RF powering at different positions on the coupler,which were not perfectly conditioned by the coupler pro- cessing. The outgassing was not substantial and was sta-bilized within 20 minutes in both cases.
V. INFLUENCE OF THERMAL CYCLES
The cryomodule was warmed up a couple of times.Itwas important to check whether the contaminations wererepopulated after thermal cycles especially because of thecross-contamination mentioned before,The small thermal cycle of the cavities from 2 K to 50K did not influence the contamination level of the couplerat cold. Even after a huge thermal cycle, in which thecavity temperature reached 293 K and was kept therefor one month, the conditioning of the coupler requiredshorter time (a few hours) than at the first conditioning(around 77 h in total). On the other hand, the cavitymultipacting was always redistributed, and it took almostthe same time (around 6 h) to finish conditioning as afterthe first cooling down.The beam vacuum was always kept below 10 − mbarduring these thermal cycles. When the cavity packagewas warming up, we observed a substantial outgassingfrom around 20 K as shown in Fig. 7. The outgassingtemperature and pressure imply that nitrogen would bethe dominant vapor element [22]. A similar outgassingalways occurred in all the thermal cycles, and proba-bly caused redistribution of the gas molecules over thecavity surface. As the surface area of the couplers issmaller than the surface of the cavities, for capturing gasmolecules, the recontamination of the couplers were notsubstantial compared to the repopulation of the cavitymultipacting. T [K]5 10 15 20 25 30 35 40 45 50 v a c uu m [ m ba r ] -9 -8 -7 -6 -5 -4 FIG. 7: Outgassing during a thermal cycle when the temper-ature of a cavity increased from 4.2 K to room temperature.
From these results, we can conclude that the first cou-pler conditioning at warm is time consuming but the fol-lowing conditionings after thermal cycles take substan-tially shorter time. Although the effect of transportationhas not been tested yet, we assume that the coupler con-ditioning time in the ESS tunnel may also be reducedthanks to the tests in the FREIA laboratory if the beamvacuum is kept under 10 − mbar after our tests. The cav-ity conditioning must be carried out every time after athermal cycle and it takes 6 h for each cavity. These areimportant feedbacks to the ESS project for optimizingconditioning procedure during RF commissioning. VI. CONCLUSION
The conditioning process of the DSRs for the ESSproject was studied in a prototype cryomodule. We de-veloped a careful and thorough conditioning procedureto safely operate the cavities. It was found that the cou-pler conditioning at warm with cross-contamination wasthe bottleneck in the total processing time. We proposehow to improve the warm conditioning by adding a sec-ond vacuum pump and do simultaneous conditioning ofboth cavity packages with an extra power station. Thecold conditioning of the coupler and the cavity did not show cross-contamination and were processed smoothly.The multipacting barriers found for the cavities were con-sistent with the vertical and horizontal tests with someremarks on the possible influence from the power coupler.After a thermal cycle both cavity and coupler condition-ing must be repeated, but the coupler conditioning re-quires substantially shorter periods. This study was thefirst experimental validation of the DSR technology andwill provide a standard method for the ESS project andsimilar future projects.
Acknowledgements
We would like to express our sincere gratitude to IPNOrsay and ESS for their valuable help and discussion onthe prototype cryomodule testing. We also thank all thetechnical staff at the FREIA laboratory. A special thanksgoes to Volker Ziemann for the useful discussion. Thisproject has received funding from the European Union’sHorizon 2020 Research and Innovation program underGrant Agreement No 730871. [1] J. Delayen, Nuclear Instruments and Methods in PhysicsResearch Section B: Beam Interactions with Mate-rials and Atoms , 892 (1989), ISSN 0168-583X, URL .[2] K. W. Shepard, P. N. Ostroumov, and J. R. Delayen,Phys. Rev. ST Accel. Beams , 080101 (2003), URL https://link.aps.org/doi/10.1103/PhysRevSTAB.6.080101 .[3] K. Shepard, M. Kelly, J. Fuerst, M. Kedzie,and Z. Conway, Physica C: Superconductivity ,205 (2006), ISSN 0921-4534, proceedings of the12th International Workshop on RF Superconduc-tivity, URL .[4] G. Olry, J.-L. Biarrotte, S. Blivet, S. Bousson, C. Joly,T. Junquera, J. Lesrel, A. Mueller, H. Saugnac,P. Szott, et al., Physica C: Superconductivity ,201 (2006), ISSN 0921-4534, proceedings of the12th International Workshop on RF Superconduc-tivity, URL .[5] C. S. Hopper and J. R. Delayen, Phys. Rev. ST Accel.Beams , 102001 (2013), URL https://link.aps.org/doi/10.1103/PhysRevSTAB.16.102001 .[6] H. Li, J.-P. Dai, P. Sha, Q.-Y. Wang, H. Huang, L.-H.Li, J. Zhang, Q. Ma, H.-Y. Lin, Y. Sun, et al., ChinesePhysics C , 077008 (2014), URL https://doi.org/10.1088%2F1674-1137%2F38%2F7%2F077008 .[7] A. Facco, Superconductor Science and Technology , 023002 (2016), URL https://doi.org/10.1088%2F1361-6668%2F30%2F2%2F023002 .[8] R. Garoby, A. Vergara, H. Danared, I. Alonso, E. Bar-gallo, B. Cheymol, C. Darve, M. Eshraqi, H. Hassan-zadegan, A. Jansson, et al., Physica Scripta , 014001 (2017), URL https://doi.org/10.1088%2F1402-4896%2Faa9bff .[9] H. Li, V. Goryashko, A. Bhattacharyya, R. S. Kern,L. Hermansson, R. Ruber, D. Dancila, and G. Olry, Pro-ceedings of SRF2015, Whistler, BC, Canada TUPB083 ,791 (2015), URL .[10] H. Li, V. Goryashko, and R. Ruber, Tech. Rep., Upp-sala University, Department of Physics and Astronomy(2015).[11] H. Li, V. Goryashko, A. Bhattacharyya, and R. Ruber,Tech. Rep. 2016/01, Uppsala University, Department ofPhysics and Astronomy (2016).[12] H. Li, M. Jobs, R. S. Kern, V. Goryashko, L. Hermans-son, A. Bhattacharyya, T. Lofnes, K. Gajewski, K. Frans-son, and R. Ruber, Nuclear Instruments and Methods inPhysics Research Section A: Accelerators, Spectrometers,Detectors and Associated Equipment , 63 (2019),ISSN 0168-9002, URL .[13] H. Li, R. Santiago-Kern, M. Jobs, A. Bhattacharyya,V. Goryashko, L. Hermansson, K. Gajewski, T. Lofnes,K. Fransson, and R. Ruber, Tech. Rep. 2017/10, UppsalaUniversity, FREIA (2017).[14] P. Duthil, D. Reynet, G. Olry, S. Brault, P. Duch-esne, N. Gandolfo, E. Rampnoux, D. Longuev-ergne, M. Pierens, F. Chatelet, et al., Proceed-ings of HB2013, Malmo, Sweden
WEAM4Y01 ,416 (2016), URL http://accelconf.web.cern.ch/accelconf/hb2016/papers/weam4y01.pdf .[15] A. Miyazaki, K. Fransson, K. Gajewski, L. Hermansson,H. Li, R. Ruber, R. Santiago Kern, and R. Wedberg,
Conditioning Experience of the ESS Spoke CryomodulePrototype , in (2019), p. THP058. [16] R. Santiago-Kern, H. Li, A. Miyazaki, K. Gajewski,L. Hermansson, K. Fransson, T. Lofnes, and R. Ruber,Tech. Rep. 2019/05, Uppsala University, FREIA (2019).[17] H. Li, A. Miyazaki, R. Santiago-Kern, L. Hermans-son, T. Lofnes, K. Gajewski, K. Fransson, R. Wedberg,and R. Ruber, Tech. Rep. 2019/08, Uppsala University,FREIA (2019).[18] M. Jobs, K. Gajewski, V. Goryashko, H. Li, R. Ruber,and R. Wedberg, Proceedings of IPAC2017, Copen-hagen, Denmark
THPIK090 , 4296 (2017), URL http://accelconf.web.cern.ch/AccelConf/ipac2017/ papers/thpik090.pdf .[19] R. Ruber, R. Wedberg, T. Peterson, D. Dancila,L. Hoang, and J. Eriksson, Tech. Rep. 2019/06, UppsalaUniversity, FREIA (2019).[20] S. Peggs,
ESS Technical Design Report (2013).[21] D. Longuevergne private communication.[22] R. E. Honig and H. O. Hook,