PLASMA PROCESSING R&D FOR THE SNS SUPERCONDUCTING LINAC RF CAVITIES*

Transcription

1 Proceedings of SRF13, Paris, France TUP57 PLASMA PROCESSING R&D FOR THE SNS SUPERCONDUCTING LINAC RF CAVITIES* M. Doleans, R. Afanador, J. Ball, W. Blokland, M. Crofford, B. Degraff, D. Douglas, B. Hannah, M. Howell, S-H. Ki, S-W. Lee, C. McMahan, J. Saunders, P. V. Tyagi, ORNL, Oak Ridge, TN 37831, USA Abstract The Spallation Neutron Source routinely operates with a proton bea power of 1 MW on its production target. A plan to reach the design 1.4 MW within a few years is in place [1] and relies on increasing the ion bea current, pulse length and bea energy in the linac. The increase in bea energy fro the present 93 MeV to 1 GeV will require an increase of approxiately 15% in the accelerating gradient of the superconducting linac highbeta cryoodules. In-situ processing was identified as a proising technique to reduce electron activity in the SNS superconducting cavities and increase their accelerating gradient []. R&D on processing ais at deploying the new in-situ technique in the linac tunnel by 16. Overall plan and current status of the processing R&D will be presented. INTRODUCTION Mediu-beta (MB) superconducting cavities at the SNS operate above design gradients while high-beta (HB) cavities operate below design gradients, ainly due to electron activities and in particular due to field eission. The electrons eitted fro the surface are accelerated by the RF electric field before striking and heating the Niobiu surface of the resonators leading to theral instabilities, ostly in the cavity end-groups. As explained in [], the accelerating gradients that can be run stably in the superconducting linac (SCL) at the SNS strongly depends on the repetition rate. The higher the repetition rate, the lower the accelerating gradients before reaching theral instability in the cryoodules. At 1 Hz repetition rate, and with cavities powered individually, the average stable accelerating gradient for the HB section is about 3% higher than at the noinal 6 Hz repetition rate with all the cavities powered siultaneously. A 15% increase of the accelerating gradients of the SCL HB cryoodules is being pursued using in-situ processing. A siilar idea is being considered at JLAB to benefit CEBAF operation [3]. Figure 1: Strategy for the developent of processing technique at the SNS. *This work was supported by SNS through UT-Battelle, LLC, under contract DE-AC5-OR75 for the U.S. DOE 551

2 TUP57 Proceedings of SRF13, Paris, France Previous studies [] and RGA easureent during theral cycling of cryoodules have shown that residual containation, for exaple fro hydrocarbons, is a source of field eission in the SNS SCL resonators. In-situ processing ais at reoving residual surface containation, particularly hydrocarbons fro the HB cavity inner surfaces. No etching technique like in [4] is being pursued at this point in tie at the SNS. Also, no electron cyclotron resonance is used to excite the as in [5] because it wouldn t be suitable for in-situ processing of the SNS cryoodules. A icrowave discharge in an SRF cavity can be generated at roo teperature or at cryogenic teperature. Because cold surfaces can cryo-pup the volatiles copounds desorbed fro the surface, processing at roo teperature is preferred. Because deploying a new technique in an operating superconducting linac presents inherent risks, a progressive strategy in four phases has been devised and is shown in Figure 1: Plasa processing studies with 3-cell and 6-cell cavities Cold-test of processed cavities Plasa processing of an offline cryoodule Deployent of in-situ processing in the SNS linac tunnel The processing of an offline cryoodule will be facilitated by the fact that the SNS has recently built a spare HB cryoodule [6]. state aplifier, a circulator and a load capable of supporting full reflection, and a crate for control of the RF and recording of all the instruents. The aplifier has a center frequency of 85 MHz and a +-5 MHz tuning range. Instruentation connected to the crate includes vacuu gauges, RF generator and power eters, video caeras, RGA, optical spectroeter and therocouples. Most of the data acquisition has been done through Labview and post-processing is done using DIAde. DIAde is an ideal data visualization tool for this type of R&D because it easily handles heterogeneous types of data, it has a high level script language and can be used to replay processing studies with synchronization of all the instruent signals. An exaple of such a sequence is illustrated in Figure 3. Figure 3: Exaple of a processing sequence postprocessed and visualized in DIAde. Plasa Ignition To ignite a in the cavity, a flow of neutral gas is created through the cavity volue and one of the fundaental passband ode is excited until a icrowave discharge occurs. The ode frequency and field aplitude in each cell of a ulticell cavity is given by [1] Figure : Plasa processing in a 6-cell beta=.81 cavity at the SNS. PLASMA PROCESSING AT THE SNS The first phase of the processing R&D is ongoing. Preparatory work for the second phase has started and cold-test of processed cavities are planned in FY14 using the new vertical test area (VTA) and horizontal test apparatus (HTA) [7,8]. The VTA and HTA use the new Cryogenic Test Facility (CTF) [9]. Plasa Processing Station A processing station was developed during the first phase of the R&D. This station is ounted on a ovable rack and is shown in Figure. In its present configuration it has two RF generators, a 5 W solid 55 j 1 k1 cos N N sin j 1/ N. (1) N Where the zero index refers to the frequency of a singlecell cavity, the index refers to the th ode, and the index j runs fro 1 to N, N being the nuber of cells. The fundaental ode resonant frequencies and onaxis field profiles for the 3-cell MB cavity and several 6- cell HB cavities used in the R&D have been easured and closely follow Equation 1. As expected, all cavities had a few percents deviation of their field intensities with respect to the ideal odel.

3 Proceedings of SRF13, Paris, France TUP57 Figure 4: (left) on-axis field easureent for the six fundaental passband odes a SNS HB 6-cell cavity. (right) atrix plot of the ideal noralized field aplitudes. As an exaple, the on-axis field profile easureent for the six fundaental odes in a 6-cell HB cavity is shown in Figure 4. A convenient 6x6 atrix plot also shows the field aplitude for all odes and in each cell of the resonator. The ignition condition in a cavity depends on the gas specie and pressure and has siilar features to the Paschen curve for a DC discharge [11]. Ignition curves for the 3-cell ediu beta cavity are shown in Figure 5. Ignition was achieved in the cavity volue for various gases and for the three odes of the 3-cell cavity. Plasa ignition in the fundaental power coupler instead of in the cavity volue can also occur but an acceptable range of gas pressure and forward power could always be found to ignite a in the cavity. Epeak (kv/) Pressure (Torr) Figure 5: Plasa ignition in a 3-cell MB cavity as a function of the gas pressure and peak surface electric field. air pi air pi/3 air pi/3 neon pi ode neon pi/3 ode neon pi/3 ode heliu pi ode heliu pi/3 ode heliu pi/3 ode Because the aplitude of the electric field in each cell depends on which ode is being excited, the can be ignited at various locations in the resonator. As an exaple, a heliu ignited in the 6 th cell or in the nd cell of a HB cavity is shown in Figure 6. Plasa Tuning The ignited in a cavity acts as a dielectric [1] with dielectric constant less than unity. The dielectric constant depends on the density and is given by 1 f rf n. () Figure 6: Heliu in a 6-cell HB cavity in the 6 th cell (top row) and nd cell (botto row). Depending on its location, the appears closer or farther away fro the axial caeras used at each end of the resonator. 553

4 TUP57 Proceedings of SRF13, Paris, France Figure 7: (left, odel) Perturbation of the resonance frequency (a.u) due to a in each of the six cells of the cavity and for each of the six fundaental passband odes of a HB cavity. (right, experient) Measureent of the resonance odes for odes 3 to 6 without (blue) and with (green) ignited in the nd cell of the cavity. Only odes 3, 5 and 6 are shifted, while ode 4 frequency reains unchanged, as expected fro the odel (see red contour in the left plot) P When a resonant structure is loaded by a, its resonant frequency is shifted upward [13] n P (5) n 1 1 eneutral rf E E cavity dv dv At our working pressure the collisional effect of the electron on the neutral gas can be neglected (i.e. <<1). The ode characteristics (i.e. resonant frequencies and field patterns) depend in which cells the has ignited. When the loading effect is treated as a perturbation, the description of the ode characteristics closely resebles the one used for field flatness tuning of a ulticell cavity [1]. For exaple if one writes the perturbation atrix due to the as 1 P N (3) (4) one can write the perturbation effect on the resonant frequencies and field aplitudes as 554 n To illustrate this effect, a atrix plot showing the shift in the ode resonant frequencies caused by a in each of the cells of a cavity is shown in Figure 7. To check the basic validity of this odel, a was ignited in the nd cell of a 6-cell HB cavity by using the nd ode. Resonances of ode 3 to 6 were onitored siultaneously through the field-probe signal. As expected fro the odel, the resonance frequency for odes 3, 5 and 6 are all shifted upward while the one for ode 4 is left unchanged. This is because the 4 th ode has no stored energy in the nd cell. As shown in Equation 5, the relative field aplitude in the cells of a resonator is also affected by the ignition of a. Figure 8 shows the perturbation on the aplitude of the electric field, for all odes and in each cell, due to the ignition of a at different locations in the cavity. The odelling of the perturbation fro a in the resonator provides relevant inforation to tune the. To increase the density, the RF generator frequency should be raised as to track the upward shift of the ode resonances. It should be noted that when this is done, the increase in the density then leads to a larger shift in the resonance frequency and the processes can be iterated to further increase the density. To ignite the in ultiple cells, the variation of the field aplitude as illustrated in Figure 8 can be utilized. n

5 Proceedings of SRF13, Paris, France TUP57 Figure 8: Perturbation of the field aplitude for all fundaental passband odes and in each cell of 6-cell cavity due to the presence of a at various location in the resonator. Results for a ignited in cell 1, or 3 are presented. Results for cells 4, 5 and 6 can found by irror syetry. For exaple, one sees that the 5 th ode is adequate to ignite a in the two end cells of the resonator. For this ode, when the ignites in one end cell, the field aplitude increases in the opposite end cell of the cavity such that the ignition level in that cell can be reached. Siilarly, the 1 st ode is adequate to ignite a in the two center-cells. Plasa Cheistry and Cleaning Plasa cheistry is a vast and rich topic [14]. The echaniss of the -cheical processes can be quite coplex because the is inherently a highly reactive syste with ulti-coponents cheically active. All the charged particles, excited atos and olecules, radicals and UV photons can play an active role and lead to physical, cheical and photocheical processes. In our case, the physical energy of the ions is too sall to initiate significant physical action and the processes are doinated by cheical and photocheical effects. It is thus very iprobable to find one processing ethod to get rid of all possible residual containants on the operating SNS cavities. We have purposefully decided to focus priarily on the reoval of hydrocarbon residues fro the surface because it is believed that their cleaning could lead to significant itigation of the field eission affecting the SNS HB cavities. But, even in this case, the task of cleaning hydrocarbons fro a niobiu surface is far fro trivial since cleaning and redepositing processes happen siultaneously. When a cheically active is generated it is as critical to onitor the possible creation of new cheical bonds on the surface as it is to onitor the cleaning of the surface containants. Niobiu saples introduced in the cavity volue provide a convenient ethod to learn and study cheistry for SRF cavities. Because we decided to study the reoval of hydrocarbons, FTIR technique [15] was used to analyse the hydrocarbons on the niobiu surface of the saples. FTIR, and particularly specular reflectance ethod [16] was used because it is a fast analysis technique that doesn t require UHV or saple preparation, it is non-destructive so that the sae saples can be analysed before and after processing, and it provides cheical bond inforation about the surface containants. Of course, FTIR also has its liitations and the use of copleentary surface analysis techniques is being pursued. Plasa processing of hydrocarbon was attepted using a Ne-O -H ixture. Neon was used as a support gas because it was found to create very stable discharges during tuning studies. Oxygen is used as a cleaning agent because it can oxidize the surface carbon to CO and CO, which easily desorb. And it can also react with the surface hydrogen and desorb in water and hydroxyl for [17-19]. However, it can also attach during the process []. For this reason, hydrogen was added to the ixture. Surface carbon is hydrogenated to hydrocarbon olecules and oxygen can be reoved by creating and desorbing water and hydroxyl olecules [1]. Four Niobiu saples (flat disks of 5 diaeter and 1 thickness) were containated using a black peranent arker which contains long hydrocarbon chains such as n-propanol and n-butanol []. The saples were introduced in the 3-cell MB cavity. Two were positioned close in the iris region of the end cells, and the other two far away fro any electric field in the end flanges region as shown in Figure 9. FTIR spectra for all saples before processing are shown in Figure 1. The coplexity of the hydrocarbon coposition of the arker is evident fro the broadness and ultiplicity of the peaks in the FTIR spectra. 555

6 TUP57 Proceedings of SRF13, Paris, France were detected). But, copleentary analysis techniques such as XPS or SIMS [15] should be perfored on the processed saples to confir this point. It is believed that the reaining traces of containation on saple and 3 could be reoved using additional processing. Figure 9: Four Nb saples before introduction into the 3- cell MB cavity. The saples were purposefully containated to study the cleaning of hydrocarbon chains by a. Figure 11: Nb saples after processing in Ne-O - H ixture. The saples positioned near the end cells of the cavity were cleaned while the saples located near the end flanges reain basically unchanged. A using a Ne-O -H ixture was ignited using the second ode and tuned up such that a was present in both end-cells of the cavity volue. As expected, increase in the RGA signals for asses 18, 8 and 44 were observed when the was ignited corresponding to reoval of hydrocarbons fro the saples in the for of water, carbon onoxide and carbon dioxide. The was left on for a few hours. The four saples were then reoved fro the cavity volue for inspection. Figure 1: FTIR specular reflectance spectra pectra for the four Niobiu saples after processing. CONCLUSION Plasa processing R&D at the SNS is on-going and ais at providing an in-situ technique to reove residual containation for the SC cavities operated in the linac. Results fro the first phase of the R&D are proising and the second phase of the R&D will start in FY14 and include cold-test of processed cavities. Figure 1: FTIR specular reflectance spectra for the four Niobiu saples before processing. Iages of the saples after processing are shown in Figure 11, and the corresponding FTIR spectra are shown in Figure 1. The two saples positioned close to the in the end-cells were significantly cleaned by the, while the other two saples located in the end-flanges outside of the do not appear significantly changed. Furtherore, the FTIR spectra reveal that all the types of hydrocarbon bonds were successfully processed and that no new cheical bonds were fored on the saple surface (i.e. no new peaks 556 ACKNOWLEDGMENT We would like to thank John Maosser fro Thoas Jefferson National Accelerator Facility for useful discussions. REFERENCES [1] Neutron Sciences Directorate Five year plan 1-16, August 1, SNS-NSCD-EXE-PN-1, R, Oak Ridge National Laboratory. [] S-H Ki et al., R&D Status for In-Situ Plasa Surface Cleaning of SRF Cavities at Spallation Neutron Source, Proceedings of 11 Particle Accelerator Conference, New York, NY, USA. [3] J. Maosser et al., Large-volue Resonant Microwave Discharge for Plasa Cleaning of a

7 CEBAF 5-cell SRF Cavity, Proceedings of IPAC1, New Orleans, Louisiana, USA. [4] M. Raskovic, Plasa treatent of bulk niobiu surface for superconducting rf cavities: Optiization of the experiental conditions on flat saples PRSTAB 13, 111 (1). [5] G. Wu et al., Plasa cleaning: towards high yield cavity processing, TFSRF 8, JLAB. [6] M. Howell et al., The First ASME Code Staped Cryoodule at SNS, IPAC 1, New Orleans, LA, May 1, p. 465 (1). [7] S.H. Ki et al., The Status of Superconducting Linac and SRF Activities at the SNS, Proceedings of the 16 th International Conference on RF Superconductivity, Paris, France, 13. [8] J. Saunders et al, Status of SRF Facilities at SNS, IPAC 1, New Orleans, LA, May 1, p. 471 (1). [9] M. Howell et al, Status of Spallation Neutron Source Cryogenic Test Facility (CTF), subitted to PAC 13, Pasadena, CA, Septeber 13. [1] H. Padasee, J. Knobloch and T. Hays, RF Superconductivity for Accelerators Wiley series in bea physics and accelerator technology, ISBN [11] S. C. Brown, Introduction to Electrical Discharges in Gases Wiley series in physics. [1] S. Bastian, Notes on electroagnetic waves in a, National Radio Astronoy Obsevatory, Charlottesville, VA, 5. [13] S. C. Brown Microwave Studies of Gas Discharge Plasas, P/387 USA. [14] A. Fridan, Plasa Cheistry Cabridge University Press, ISBN [15] J. Vickeran and I. Gilore, Surface Analysis - The Principal Techniques J. Wiley & Sons, Ltd. DOI: 1.1/ [16].Setnicka, FT-IR Reflection Techniques. [17] R. E. Kirby et al., Linear Collider Collaboration Tech Notes, LCC-75, Noveberr 1, Stanford Linear Accelerator Center, Stanford, CA. [18] A. Belkind et al., Oil reoval fro etals by linear ulti-orifice hollow cathode, Surface and Coatings Technology (1995) [19] H. Li et al., An in situ XPS study of oxygen cleaning of aluinu surfaces, Surface and Coatings Technology 9 (1997) I [] A. Vesel, Modification of polystyrene with a highly reactive cold oxygen, Surface & Coatings Technology 5 (1) [1] P. Kruger, R. Knes and J. Friedrich, Surface cleaning by -enhanced desorption of containants (PEDC), Surface and Coatings Technology 11 (1999) [] C. Moody, Black writing ink analysis by direct infusion electrospray ass spectroetry, M.S. Thesis, University of central Florida, (1). Proceedings of SRF13, Paris, France TUP57 557