High gradient Superconducting Cavity Development for FFAGs

1. High gradient Superconducting Cavity Development for FFAGs Sergey V. Kutsaev, Zachary A. Conway, Peter N. Ostroumov C. Johnstoneand R. Ford Cyclotrons’13 September 23, 2013 Vancouver, BC, Canada

2. FFAG is typically configured to accelerate a large momentum admittance within a modest magnet aperture. 1 GeV CW FFAG for C Therapy C. Johnstone – Trinity College, 2011 3 to 10 GeV muon double beam FFAG T. Planche - Nufact09 They are brilliant designs, as long as you don’t want too much beam current… P. McIntyre, FFAG13 Confusion about the new breed of nonlinear nsFFAGs with constant tune (above and PAMELA, for example) and EMMA (swept tune) FFAG13 TRIUMF

3. Advanced design and simulation of anIsochronous 250-1000 MeVNonscaling FFAG 2m General Parameters of an initial 0. 250 – 1 GeV non-scaling, near-isochronous FFAG lattice design Clockwise: Matematica: Ring tune, deviation from isochronous orbit (%), and radius vs. momentum • Comments and further work • Tracking results indicate ~50-100 mm-mr; relatively insensitive to errors • Low losses

4. FFAG cavity

5. F-D quads control betatron motion Uniform gradient in each channel: excellent linear dynamics. 5.5 5.0 4.5 4.0 We can lock nx, ny to any desired operating point. BTC quads are tuned in 2 x 5 families. Sextupole correctors at exit of each BTC are tuned in 2 x 6 families. First 2 turns each have dedicated families so that they can be tuned first for rational commissioning. 0.0 0.5 1.0 1.5 2.0 2.5 P. McIntyre, FFAG13, “ Strong-focusing cyclotron” FFAG13 TRIUMF

6. Now look at effects of synchrobetatron and space charge with 10 mA at extraction: Move tunes near integer fraction resonances to observe growth of islands After two turns … bunch is lost by 20 MeV 1/3 order integer effect 1/5 order integer effect 1/5-order islands stay clumped, 1/3-order islands are being driven. Likely driving term is edge fields of sectors (6-fold sector geometry). We are evaluating use of sextupoles at sector edges to suppress growth. FFAG13 TRIUMF

7. Synchrobetatron/space charge in longitudinal phase space: Tunes again moved to approach resonances, but retaining transmission through lattice Injection extraction Phase width grows x5 at extraction FFAG13 TRIUMF

8. Simulations of a proton FFAG with space charge FFAG’12, Osaka, Japan Dr. Suzie Sheehy, ASTeC/STFC

9. Overview • Motivation • 4-cell 1GeV FFAG • Single energy simulations • Serpentine acceleration with SC • 6-cell 1GeV FFAG simulations • Single energy simulations • Serpentine acceleration with SC • Two discussion points: • Cost to build an FFAG vslinac • Terminology • Summary Dr. Suzie Sheehy, ASTeC/STFC

10. 4-cell Lattice design F Total tune variation = 0.12 (H) / 0.56 (V) D C. Johnstone et al, AIP conference proceedings 1299, 1, 682-687 (2010) Dr. Suzie Sheehy, ASTeC/STFC

11. Serpentine channel acceleration • Using Carol’s 4-cell 1 GeV FFAG ring • 330MeV-1GeV • Serpentine acceleration is possible due to 3% time of flight variation Using single particle tracking: 10 MV/turn 12 MV/turn 15 MV/turn From IPAC’11, S. L. Sheehy, WEPS088 Dr. Suzie Sheehy, ASTeC/STFC

12. Setting up the simulation in OPAL • Beam matching at 300 MeV • Single energy tracking at realistic intensities shows no emittance growth. • At very high intensities emittance grows, as expected. • (See HB2012 MOP258) Single bunch 10 π mm mrad in horiz/vertical Length ~ 4% of ring circumference 50 turns at 300 MeV with no acceleration Current is average of full turn Dr. Suzie Sheehy, ASTeC/STFC

13. Setting up the simulation • Harmonic number = 1 (for now) • Space charge ‘off’ – track bunch of 2000 particles with parabolic profile Energy (MeV) Cavity phase (deg) Issue: OPAL outputs after set num of tracking steps NOT azimuthal position ie. not a full ‘turn’ if not perfectly on phase! In the newer version of OPAL (1.1.9) I can use ‘probe’ element to get ‘screen-like’ physical position readout (in process of updating on SCARF). For now – the following simulations were run overnight on my quad core desktop! Dr. Suzie Sheehy, ASTeC/STFC

14. Turn-by-turn orbit with bunch • Try a short & small 1 π mm mrad beam to see what happens • Using ‘probe’ element to get turn-by-turn at 0 degree position, 200 particles Momentum [βγ] Acceleration! WithVRF/turn= 22 MV Dr. Suzie Sheehy, ASTeC/STFC X Position [mm]

15. With space charge ‘on’ • 1 pi mm mradbunch • Tracked for 30 turns in serpentine channel • Vary (average) beam current up to 10 mA • 10 MW beam at 1 GeV • NOTE: VERY SHORT BEAM! • Total beam length approx. 1cm so peak current = 22A! • Lots of space charge!! Acceleration with 3 cavities VRF/turn= 22 MV Dr. Suzie Sheehy, ASTeC/STFC

16. Longitudinal evolution ΔE [MeV] 1 2 5 10 28 15 20 25 Δt [ns] Dr. Suzie Sheehy, ASTeC/STFC

17. Horizontal posvs time 1 2 5 10 Δx [mm] 20 25 26 15 Δt [ns] Dr. Suzie Sheehy, ASTeC/STFC

18. Vertical posvs time 1 2 5 10 Δz [mm] 20 25 26 15 Δt [ns] Dr. Suzie Sheehy, ASTeC/STFC

19. (NB as discussed before these are per ‘turn’ wrt time so not ‘real’ turns) Emittance evolution NOTE DIFFERENT SCALES!! 300 MeV – 30 turn serp. Accel. 300 MeV – no acceleration horizontal Question: why does vertical emit grow even with no space charge? vertical longitudinal Dr. Suzie Sheehy, ASTeC/STFC

20. 6-cell 1GeV Ring • Initially I had issues modeling this ring as short magnets were not well-reproduced with the interpolation in OPAL • I now have a much finer field map! (thanks to K. Makino) • The main driver behind this design was to achieve better isochronicity but the vertical tune variation also improved. • Total tune variation = 0.42 (H) / 0.234 (V) • Lower B field (cf 2.35T in 4-cell design) • Larger radius Field profiles from C. Johnstone, IPAC’12, THPPR063 Dr. Suzie Sheehy, ASTeC/STFC

21. Single energy with space charge • No acceleration, 10 pi beam • 50 turns with space charge • Short beam (0.04% ring) • No acceleration, 10 pi beam • 50 turns with space charge • Longer beam (4% ring) Dr. Suzie Sheehy, ASTeC/STFC

22. Serpentine channel • Much smaller TOF variation than 4-cell version • Required RF frequency & energy gain per turn to open up the serpentine channel (δE) can be estimated by total energy gain required is ΔE and ω =2πFRF In this case, we expect δE~ 3.7 MV/turn. (A few quick simulations confirm this) 4MV/turn ! If we make it ‘isochronous enough’, perhaps we can use the serpentine channel without superconducting cavities? Dr. Suzie Sheehy, ASTeC/STFC

23. Acceleration & space charge • Use ‘reasonable’ 8MV/turn (equiv. 4 PSI-type cavities) • Beam matched at 330 MeV • Same emittance & length beam as in 6-cell simulations Dr. Suzie Sheehy, ASTeC/STFC

24. Outline • Motivation and Background • Next generation ultra-compact, high-energy fixed field accelerators • Medical, security, energy applications • CW FFAGs ; i.e. strong-focusing cyclotrons • Relativistic energies: ~200 MeV – 1 GeV • Ultra-compact • Constant machine tunes (optimized gradients) • High mA currents (low losses) • These machines require high gradient acceleration; SCRF required for • Compactness • Low extraction losses • Large horizontal aperture of the FFAG, like the cyclotron, is a challenging problem for SCRF design FFAG cavity

25. NEXT-generation CW high-energy Fixed-field Compact Accelerators • Reverse gradient required for vertical envelope • Isochronous or CW (serpentine channel relaxes tolerances) • Stable tune, large energy range • The footprint of CW FFAG accelerators is decreasing rapidly Machine tunes: r ~1.4 z ~0.8 – factor of ~4 > than compact cyclotron

26. MAGNETS and modeling < 3 m < 5 m One straight section occupied by RF cavities and injection/extraction in the other FFAG Horizontal / Vertical Stable beam area @200 MeV FFAG Horizontal Stable beam area@1000 MeV vs. DA of 800 MeV Daealus cyclotron* cyclotron cyclotron 10 mm x 9 mr V= 90π mm-mr norm=62π mm-mr Tracked: 130 mm x 165 mr  = 21450π mm-mr norm = 38820π mm-mr 80 mm x 293 mr H= 23,440π mm-mr norm = 16,059π mm-mr F. Meot, et. al., Proc. IPAC2012 Tracking: Horizontal – 1 cm steps, Vertical – 1 mm steps *FFAG vert. stable area at aperture limits.

27. Acceleration Gradient required for low-loss extraction Reference radius in center of straight for the energy orbits preceding extraction. For an accelerating gradient of ~20 MV/m orbits are sufficiently separated for a “clean” (beam size: 1.14 cm; =10 mm-mr normalized) or low-loss extraction through a septum magnet. For 20 MV/turn, and a 2m straight section, we require 10 MV/m – implies a SCRF cryomodule – in order to achieve extraction with manageable shielding, radiation levels, and activation. This requirement drove the design of the high-energy stage.

28. Design specifications • Large horizontal beam aperture of 50 cm • Cavity should operate at 150 or 200 MHz (harmonic of the revolution frequency) • Should provide at least 5 MV for proton beam with energies 200 – 900 MeV • Peak magnetic field should be no more than 160 mT (preferably, 120 mT or less) • Peak electric field should be minimized • Cavity dimensions should be minimized FFAG cavity

29. Cavity options • Half-wave resonator H-Resonators Beam trajectories HWR is very dependent on particle velocity Can’t be used efficiently for such a wide range of particle energies Dimensions are very large as is peak magnetic field on the electrode edge FFAG cavity

30. Rectangular Cavity • Rectangular cavity operating at H101 mode has electric field concentrated in the center of the wall • To concentrate electric field at beam aperture, we introduced tapers • To reduce peak magnetic field the blending was introduced Beam direction W H L FFAG cavity

31. Gap and Frequency Optimization • The voltage at 160 mT maximum field dependence on gap length was calculated for cavities with different frequencies and lengths Beam Energy = 200 MeV Voltage in the center of the aperture Peak magnetic field = 160 mT 150 MHz 1.5 m structure has a potentially higher possible voltage or lower peak magnetic field at 5 MV 200 MHz structure is more compact FFAG cavity

32. Cavity shape optimization • A taper was introduced to distribute the magnetic field over a larger volume keeping the electric field concentrated around the beam aperture • Such a cavity design has smaller dimensions for the same volume • All edges were rounded and improved reentrant nose shape reduced the peak magnetic field by more than 15% and the transverse dimensions by more than 10 cm • Final study was an elliptical cell shape where the magnetic field varies along the cavity wall such that there are no stable electron trajectories and multipacting is inhibited FFAG cavity

33. Comparison of different cavity geometries • Rectangular cavity is better than elliptical by all parameters except multipactor resistivity Parameters of the different cavity designs. The RF performance degrades at higher frequencies. FFAG cavity

34. RF input coupler design • As 1 mA beam is accelerated by 4 cavities from 200 to 900 MeV, each cavity requires about 175 kW of power • One of the options is to attach 2 100kW couplers to the cavity 80K 126K 4K Heat Flows: To 4K = 9.8W To 60K = 92.0W From 300K = 18.8W 300K 133K ANSYS estimations show no significant overheating FFAG cavity

35. Magnetic power coupling and mechanical design • External Q-factor should be ~ 1.9*106 • Preliminary results predict ~1.1mm Nb and ~0.6mm SS deformation at magnetic field area The complete mechanical design: 1 – niobium shell, 2 – RF ports, 3- extra ports, 4 – frequency tuning, 5 – steel jacket, 6 – rails FFAG cavity

36. Summary • Two options of 200 MHz cavities were studied: "rectangular" and " elliptical". The "elliptical" option has been introduced to avoid multipacting (MP) problem. The latter can be problem for rectangular cavity • However, the rectangular option provides better parameters (peak fields, dimensionsetc) • 100 kW RF coupler has been designed. More studies are needed to minimize cryogenic load. • Initial mechanical design was created and structural analysis has been performed. Both cavities can be made structurally stable. FFAG cavity