1. DC Gun SRF Linac UV FEL Transport Line Dump IR Wiggler Bunching Chicane The Jlab UV FEL Driver ERL

2. Delivery of appropriately configured beam to FEL Transverse, longitudinal phase space management beam size/divergence: optical mode overlap Bunch compression: high peak current at FEL Preservation of beam quality space charge, wake/collective effects, CSR Power recovery from exhaust beam Transverse, longitudinal matching Collective effect/instability control space charge BBU, FEL/RF interaction Loss (halo) management Design Requirements

3. Parameters (Achieved) ParameterIRUV Energy (MeV)88-165135 I ave (mA)9.12 Q bunch (pC)13560  N transverse/longitudinal (mm-mrad/keV-psec) 8/755/50   p/p,  l (fsec) 0.4%, 1600.4%, 100 I peak (A)400250 FEL repetition rate (MHz) (cavity fundamental 4.6875) 0.586-751.172-18.75  FEL 2.5%0.8%  E full after FEL ~15%~7%

4. Relevant Phenomena Space charge (transverse, longitudinal) Inject long (2.5 psec/1.33 o rms), low momentum spread (~¼%) bunch (LSC) BBU CSR Compress high charge bunch => potentially degrade beam quality (and get clear signature of short bunch) & put power where you don’t want it…signature of short bunch Other wake, impedance effects RF heating, resistive wall

5. DC Gun SRF Linac UV FEL Transport Line Dump IR Wiggler Bunching Chicane Design Concept: add-on to IR Upgrade retain beam dynamics solution from IR – Use same modular approach (and same optics modules) as in IR side of machine divert beam to UV FEL with minimal operational modification

6. Design Solution gun injector merger linac Transverse match: linac to arcBates bend recovery Bates bend UV bypass transport grafted onto Bates bend Betatron match to wiggler wiggler Betatron match from wiggler to recovery transport Return bypass transport to energy recovery arc Reinjection/recovery transverse match Energy recovery through linac Extraction line to dump

7. Phase Space Management Transport system is “functionally modular”: design embeds specific functions (e.g. beam formation, acceleration, transverse matching, bending, dispersion suppression, etc) within localized regions – largely avoids need for S2E analysis allows use of potentially ill-defined/poorly controlled components – demands design provide operational flexibility – requires use of beam-based methods needs extensive suite of diagnostics & controls Its a cost-performance optimization (i.e. religious) issue: pay up front for a sufficient understanding of physics, component/hardware quality, or provide operational flexibility and adequate diagnostic capability?

8. “Modules” for Phase Space Control Transverse matching – quad telescopes in nondispersed regions; decoupled from longitudinal match – Injector to linac – Linac to recirculator – Match to wiggler – Match out of wiggler to recovery transport – Reinjection match Longitudinal matching – handled in Bates bends – Path length variable over ~± RF /2 (for control of 2 nd pass RF phase) – Independent control of momentum compaction through third order (M 56, T 566, W 5666 ) and dispersion through 2 nd order (T 166, T 266 ) – Relatively decoupled from transverse match Phase space exchange (IR side only) – H/V exchange using 5 quad rotator for BBU control – Decoupled from transverse, longitudinal matching

9. Transverse Matching (Linear) Multiple quad telescopes along transport system massage H/V phase space to match to lattice acceptance – Injector to linac (4 quads) – Linac to recirculator (6 quads) – Match to wiggler (6 quads) – Match wiggler to recovery transport (6 quads) – Reinjection match (6 quads) Key points – ERLs do not have closed orbits nor do they need to be betatron stable – ERLs may not have uniquely defined “matched” Twiss envelopes – Deliberate “mismatch” (to locally stable transport) may be beneficial e.g. to manage chromatic aberrations, halo, avoid aperture constraints – Design optimization must explore parameter space to determine “best” choice – Beam envelopes and lattice Twiss parameters are **not** in general the same!

10. Operationally … Measure beam envelopes – multislit, quad scan, and/or multi-monitor emittance measurement Back-propagate results to reference point upstream of matching region Adjust quads to “match” envelopes to design values and/or acceptance of specific sections of transport system lattice Caveat: RF focusing is VERY important – dominates behavior in injector is the “observable” used to set phase on a daily basis – defines injector-to-linac match Limits tolerable gradient in first (last) cavity

11. Longitudinal Matching Space charge forces injection of a long bunch (SRF gradients are too low to preserve beam quality if bunch is short) – Must compress bunch length during/after acceleration to produce high peak current needed by FEL After lasing, beam energy spread is too large (15-20 MeV) to recover without unacceptable loss – Must energy compress (during energy recovery) to “fit” beam into dump line acceptance The manipulations needed to meet these requirements constitute the longitudinal match

12. Longitudinal Matching Schematic Longitudinal Matching for ERL-Driven FEL E  E  E  “oscillator” “amplifier” E  E  injector dump wiggler linac E 

13. Energy Recovery: Details Longitudinal Match to Wiggler Inject long, low-energy-spread bunch to avoid LSC problems – need 1-1.5 o rms with 1497 MHz RF @ 135 pC in our machine Chirp on the rising part of the RF waveform – counteracts LSC – phase set-point then determined by required momentum spread at wiggler Compress (to required order, including curvature/torsion compensation) using recirculator compactions M 56, T 566, W 5666,… Entire process generates a parallel-to-point longitudinal image from injector to wiggler

14. Longitudinal Match to Dump FEL exhaust bunch is short & has very large energy spread (10-15%) => Must energy compress during energy recovery to avoid beam loss linac during energy recovery; this defines the longitudinal match to dump – Highest energy must be phase-synchronous with (or precede) trough of RF wave-form – Transport momentum compactions must match the slope (M 56 ), curvature (T 566 ), torsion (W 5666 ),… of the RF waveform Recovered bunch centroid usually not 180 o out of phase with accelerated centroid – Not all RF power recovered, but get as close as possible (recover ahead of trough), because… – Additional forward RF power required for field control, acceleration, FEL operation; more power needed for larger phase misalignments For specific longitudinal match, energy & energy spread at dump does not depend on lasing efficiency, exhaust energy, or exhaust energy spread – Only temporal centroid and bunch length change as lasing conditions change The match constitutes a point-to-parallel image from wiggler to dump

15. Energy Compression Beam central energy drops, beam energy spread grows Recirculator energy must be matched to beam central energy to maximize acceptance Beam rotated, curved, torqued to match shape of RF waveform Maximum energy can’t exceed peak deceleration available from linac – Corollary: entire bunch must preced trough of RF waveform E t E t All e - after trough go into high- energy tail at dump E t

16. Higher Order Corrections Without nonlinear corrections, phase space becomes distorted during deceleration Curvature, torsion,… can be compensated by nonlinear adjustments – differentially move phase space regions to match gradient required for energy compression E t Required phase bite is cos -1 (1-  E FEL /E); this is >25 o at the RF fundamental for 10% exhaust energy spread, >30 o for 15% –typically need 3 rd order corrections (octupoles) –also need a few extra degrees for tails, phase errors & drifts, irreproducible & varying path lengths, etc, so that system operates reliably In this context, harmonic RF very hard to use…

17. JLab IR Demo Dump core of beam off center, even though BLMs showed edges were centered (high energy tail)

18. Longitudinal Matching Scenario Requirements on phase space: high peak current (short bunch) at FEL – bunch length compression at wiggler using quads and sextupoles to adjust compactions “small” energy spread at dump – energy compress while energy recovering – “short” RF wavelength/long bunch, large exhaust  p/p (~10%)  get slope, curvature, and torsion right (quads, sextupoles, octupoles) E  E  E  E  E  E 

19. Nonlinearity Control Validated By Measurement Figure 1: Inner sextupoles to 12726 g-cm and trim quads to -215 g Figure 2: trim quads at -185 g with same sextupoles Figure 3: trim quads at -245 g Figure 4: quads at -215, but sextupoles 3000 g below design, at 10726 g-cm Figure 5: where we left it: trim quads -215 g sextupoles at 12726 g-cm arrival  launch 

20. Injector to Wiggler Transport

21. If you do it right linac produces stable ultrashort pulses Can regularly achieve 300 fs FWHM electron pulses ~150 fsec rms

22. Injector to Reinjection Transport

23. Module Design: Injector 42 psec (FWHM) 75 (1497/20) MHz green drive laser pulse train with flexibly gated time structure DC photocathode (GaAs) gun at 325 kV Room-temp buncher providing initial longitudinal control Pair of 5-cell SRF (CEBAF) cavities at 1497 MHz accelerate to ~10 MeV Quad telescope to match to linac

24. Cathode Cesiated GaAs – Excellent performance for R&D system When charge lifetime limited, get 500 C between cesiations (50k sec, ~14 hrs at 10 mA, many days at modest current), O(10 kC) on wafer – Typically replace because we destroy wafer in an arc event, can’t get QE When (arc, emitter, vacuum,…) limited, ~few hours running – Not entirely adequate for prolonged user operations Other cathodes? – Need proof of principle for required combination of beam quality, lifetime?

25. Injector Design and Operation At highest level… – System is moderately bright & operates at moderate power – Halo & tails are significant issue – Must produce very specific beam properties to match downstream acceptance; have very limited number of free parameters to do so Issues: – Space charge & steering in front end – Deceleration by first cavity – Severe RF focusing (with coupling) – FPC/alignment steering – phasing a challenge Miniphase – Halo/tails Divots in cathode; scattered drive laser light; cathode relaxation; … (V w/ PM, MSE) (V w/ PM, BPM) (BPM) (V) Wafer 25 mm dia Active area 16 mm dia Drive laser 8 mm dia 350 kV/2.5 MV 500 kV/2.5 MV 350 kV/5 MV 500 kV/5 MV Courtesy P. Evtushenko

26. Injector Operational Challenges At highest level… – System is moderately bright & operates at moderate power – Halo & tails are issue – Must produce very specific beam properties for rest of system, and have very limited number of free parameters to do so Space charge: have to get adequate transmission through buncher – steering complicated by running drive laser off cathode axis (avoid ion back-bombardment) – solenoid must be reoptimized for each drive laser pulse length – Vacuum levels used as diagnostic (V w/ PM, MSE) (V w/ PM, BPM) (BPM) (V)

27. Injector Operational Challenges 1 st cavity – decelerates beam to ~175 keV, aggravates space charge; – E(  ) nearly constant for ±20 o around crest (phase slip) Normal & skew quad RF modes in couplers violate axial symmetry & add coupling Dipole RF mode in FPC – Steer beam in “spectrometer”, make phasing difficult – Drive head-tail emittance dilution (V w/ PM, MSE) (V w/ PM, BPM) (BPM) (V) 350 kV/1.5 MV 500 kV/1.5 MV 350 kV/2.5 MV 500 kV/2.5 MV

28. Injector Operational Challenges FPC/cavity misalignment steering ~ as big as dispersive changes in position – Phasing takes considerable care and some time – Have to back out steering using orbit measurement in linac RF focusing very severe – can make beam large/strongly divergent/convergent at end of cryounit – constrains ranges of tolerable operating phases Phasing – 4 knobs available: drive laser phase, buncher phase, 2 SRF cavity phases – Constrained by tolerable gradiants, limited number of observables (1 position at dispersed location), downstream acceptance – Typically spectrometer phase with care every few weeks; “miniphase” every few hours (V w/ PM, MSE) (V w/ PM, BPM) (BPM) (V)

29. “Miniphase” System is under-constrained, difficult to spectrometer phase with adequate resolution Phases drift out of tolerance over few hours Recover setup by 1.Set drive laser phase to put buncher at “zero crossing” (therein lies numerous tales, … or sometimes tails...) 2.Set drive laser/buncher gang phase to phase of 1 st SRF cavity by duplicating focusing (beam profile at 1 st view downstream of cryounit) 3.Set phase of 2 nd SRF cavity by recovering energy at spectrometer BPM this avoids necessity of fighting with 1 st SRF cavity… (V w/ PM, MSE) (V w/ PM, BPM) (BPM) (V)

30. Module Design: Merger Injection line: 3 dipole “Penner bend” – Achromatic (M 16, M 26 = 0) – Modest M 56 <0 providing some bunch length control Very strongly focusing in bend plane; drift-like in non-bend plane – Focusing => space charge management – drift => transverse asymmetry => bad for space charge Adequate performance at ~100 pC x several mm-mrad level hard to match across somewhat archaic design Reinjection line: achromatic chicane with geometry matched to final dipole of Penner bend – mechanically convenient (fair amount of space)

31. Numbers  =0.6 m  = 20 o normal entry/exit bend plane focal length ~  /sin  ~ 1.75 m bend-to-bend drift ~ 1.5 m dispersion in center dipole ~0.5 m M 56 ~ -0.18 m

32. Merger Issues Low charge (135 pC), low current (10 mA); beam quality preservation notionally not a problem; however… Can have dramatic variation in transverse beam properties after cryounit 4 quad telescope has extremely limited dynamic range Must match into “long” linac with limited acceptance – Matched envelopes ~10 m, upright ellipse – Have to get fairly close (halo, scraping, BBU,…) Beam quality is match sensitive (space charge) Have to iterate injector setup & match to linac until adequate performance achieved (V w/ PM, MSE) (V w/ PM, BPM) (BPM) (V)

33. Module Design: Linac Three JLab-standard (more or less) cryomodules – Outboard pair: 8 five-cell 1497 MHz cavities – Middle module: 8 seven-cell 1497 MHz cavities Triplet focusing in warm regions Relies on RF focusing in front (back) end on 1 st (2 nd ) pass Wear and tear limits full energy to ~135 MeV or so ~30 m

34. Space Charge – Esp. LSC – Down Linac Had a number of issues in linac during commissioning: – Why was the bunch “too long” at the wiggler? bunch length at wiggler “too long” even when fully “optimized” (with good longitudinal emittance out of injector) – could only get 300-400 fsec rms, needed 200 fsec – Why did the “properly tuned lattice” not fully compress the bunch? M 55 measurement showed proper injector-to-wiggler transfer function, but beam didn’t “cooperate”… minimum bunch length at “wrong” compaction – Why was the beam momentum spread asymmetric around crest? dp/p ahead of crest ~1.5 x smaller than after crest; average ~ PARMELA We blamed wakes, mis-phased cavities, fundamental design flaws, but in reality it was LSC… PARMELA simulation (C. Hernandez-Garcia) showed LSC-driven growth in correlated & uncorrelated dp/p; magnitudes consistent with observation Simulation showed uncorrelated momentum spread (which dictates compressed bunch length) tracks correlated (observable) momentum spread

35. Space-Charge Induced Degradation of Longitudinal Emittance Mechanism: self-fields cause bunch to “spread out” – Head of bunch accelerated, tail of bunch decelerated, causing correlated energy slew Ahead of crest (head at low energy, tail at high) observed momentum spread reduced After crest (head at high energy, tail at low) observed energy spread increased – “Intrinsic” momentum spread similarly aggravated (driving longer bunch) Simple estimates => imposed correlated momentum spread ~1/L b 2 and 1/r b 2 – The latter observed – bunch length clearly match-dependent – The former quickly checked…

36. Solution Additional PARMELA sims (C. Hernandez-Garcia) showed injected bunch length could be controlled by varying phase of the final injector cavity. – bunch length increased, uncorrelated momentum spread fell (but emittance increased) – reduced space charge driven effects – both correlated asymmetry across crest and uncorrelated induced momentum spread When implemented in accelerator: – final momentum spread increased from ~1% (full, ahead of crest) to ~2%; – bunch length of ~800–900 fsec FWHM reduced to ~500 fsec FWHM (now typically 350 fsec) – bunch compressed when “decorrelated” injector-to-wiggler transfer function used (“beam matched to lattice”)

37. Happek Scan

38. Key Points “Lengthen thy bunch at injection, lest space charge rise up to smite thee” (Pv. 32:1, or Hernandez-Garcia et al., Proc. FEL ’04) “best” injected emittance DOES NOT NECESSARILY produce best DELIVERED emittance! LSC effects visible with streak camera E t E t

39. Streak Camera Data from IR Upgrade -5 o -6 o 0o0o -1 o -2 o -3 o -4 o (t,E) vs. linac phase after crest (data by S. Zhang, v.g. from C. Tennant)

40. +5 o +6 o 0o0o +1 o +2 o +3 o +4 o Streak Camera Data from IR Upgrade (t,E) vs. linac phase, before crest asymmetry between + and - show effect of longitudinal space charge after 10 MeV (data by S. Zhang, v.g. from C. Tennant)

41. ±4 and ±6 degrees off crest “+” on rising, “-” on falling part of waveform L bunch consistent with dp/p and M 56 from linac to observation point dp/p(-)>dp/p(+) on “-” side there are electrons at energy higher than max out of linac distribution evolves “hot spot” on “-” side (kinematic debunching, beam slides up toward crest…) => LSC a concern… +4 o -4 o -6 o +6 o

42. Module Design: Transverse Match to Recirculator 6 quad telescope – Match  x,y,  x,y,  x,y from linac to Bates bend – Full transverse envelope match + phase advance allows management of chromatic aberrations – Choice of phase advance = aberrations destructively interfere control of turn-to-turn phase advance – BBU suppression 9 MeV beam to dump

43. Module Design: Bates Bend Requirements “delivery of appropriately configured beam to wiggler” => – betatron matched to wiggler acceptance – Compressed bunch length – Nondispersed after recirculation transport “power recovery from exhaust beam” (without loss…) => – Betatron matched to linac acceptance – Decompressed bunch length with  56 & phasing set to insure energy compression during energy recovery – Nondispersed at reinjection => Need (nonlinearly) achromatic recirculation arc with large acceptance, good betatron behavior, and variable path length/momentum compaction (through nonlinear order)

44. Bates Bend – Originally used as the basis of an energy doubling recirculator (and a demonstration of current doubling and energy recovery in the early ’80s), the MIT implementation had >8.5% momentum acceptance – JLab FEL drivers (IR Demo, IR upgrade) have both used Bates bends

45. Bates Recirculator We used the Sargent/Flanz design from MIT because it is really robust, really easy to operate (if you instrument it) and really simple (if you think about it the right way). – Good acceptance (10+%, over 30 o RF phase, good focusing properties (esp. if matching in/out uses chromatically balanced telescopes) Ancillary benefit: No simultaneously small , , l, so CSR, space charge not aggravated (parasitic crossings not an issue)

46. Module Design: Bypass to UV FEL Bates bend directs beam to IR FEL UV FEL “must” live in wiggler pit – Wiggler stand height >> beam elevation Need by-pass solution – must provide appropriate matching Transverse spot size Bunch compression Dispersion suppression – Constrained by pit location

47. Reduce bend angle in final dipole of Bates bend to direct beam toward UV wiggler pit – “short out” half of coil pack to reduce bend angle by half Complete bend when at proper transverse offset Bypass Concept Use FODO quad transport between bends to manage dispersion, control betatron envelopes – Add 1 betatron wavelength in bend plane to image/suppress dispersion – Select out-of-bend-plane phase advance to manage envelopes, aberrations (1/2 betatron wavelength works well) – Use 7 quad FODO transport – Sextupoles at high-dispersion points (at horizontally focusing quads) provide correction of 2 nd -order dispersion

48. Module Design: Match to Wiggler Similar to linac-to-arc match – 6 quad telescope (two triplets) Set transverse envelopes, phase advances

49. Module Design: Match from Wiggler Just like previous match: 6 quad telescope, set envelopes, phase advances…

50. Module Design: Return transport to recovery arc Reflection of translation from Bates bend to UV backleg

51. Module Design: Bates Bend 2 nd Bates bend geometrically identical to 1 st one Momentum compactions different (trim quad, sextupole settings) to complete longitudinal match giving energy compression during energy recovery Details to follow…

52. Module Design: Reinjection Match More or less a mirror image of linac-to-arc match Particular care on choice of phase advance needed to ensure “chromatic balance” & adequate momentum aperture – Phase advance between/across telescopes used to control chromatic aberrations – Use “extra” quads in the triplet pairs – Select Twiss parameters in arc with care 9 MeV injected beam

53. Module Design: Recovery Pass Through Linac Two issues: – Match (beam loss, halo control, tails, scraping…) – Steering (beam loss, halo control, blah blah blah) Multipass steering tricky – Best to steer 1 st pass “locally” (insofar as possible) and thread 2 nd pass through using choice of reinjection orbit – Situation rendered complex by absence of 2-pass BPMs, RF skew quad in 5-cell cavity HOM couplers

54. Multipass Orbit Correction Energy of each pass differs at any point of linac – Orbit response to steering differs Need localized observation & correction – Multipass BPMs - for high-frequency CW bunch trains separated by ½ RF period... Example: JLab IR Upgrade: orbit bump on 1 st pass of linac

55. System Layout Requirements on phase space: high peak current (short bunch) at FEL bunch length compression at wiggler using quads and sextupoles to adjust compactions “small” energy spread at dump energy compress while energy recovering “short” RF wavelength/long bunch, large exhaust  p/p (~12%)  get slope, curvature, and torsion right (quads, sextupoles, octupoles)

56. Multipass Orbit Correction – Common Transport “multiple beams” not limited to linac – some ERL geometries use single beamlines for multiple beams – E.g. CEBAF-ER

57. Module Design: Extraction Line to Recovery Dump Mirror image of injection merger Quad triplet after extraction => betatron spot control at dump face Dispersion ~ 1 m: useful diagnostic for phasing beam during energy recovery & energy compression